EFFECTS OF OVARIAN AND STRESS HORMONES ON LEARNING PROCESSES
by
Kiranjot Jhajj
B.Sc. Honours, University of Northern British Columbia, 2020

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
PSYCHOLOGY

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
August 2023
© Kiranjot Jhajj, 2023

ii
Abstract
Learning is controlled by two interacting processes, cognitive and habitual learning. How
these two systems are used while we learn in our everyday lives depends on an individual’s
context. For example, stress is one contextual factor that consistently has created a shift
toward habitual learning. In addition, there is evidence that ovarian hormones can also
influence learning processes. However, research investigating these hormonal influences has
resulted in inconsistent findings. While there is evidence that both of these contextual factors
influence learning processes, there is little research on what effects result from their
interaction. Further, while the menstrual cycle is often used to approximate ovarian hormone
levels in such studies, it is conceptualized strictly as a biological phenomenon, despite
evidence supporting its biopsychosocial characterization. Thus, the current study investigated
the individual and interactive effects of chronic stress and ovarian hormones on learning
processes, while using a biopsychosocial understanding of the menstrual cycle. Participants
(N = 32) completed a probabilistic classification learning task. They also provided salivary
measures of estradiol and progesterone, and completed measures of chronic stress and
menstrual-related attitudes and beliefs. Results revealed a trending association between
progesterone and learning processes. Further, there was an interaction between chronic stress
and estradiol in predicting learning process use. Lastly, there were significant correlations
between learning processes and various menstrual beliefs. As such, these preliminary results
revealed how ovarian hormones and chronic stress interact to influence learning processes,
and menstrual attitudes and beliefs can provide a more detailed understanding of these
effects.

iii
Table of Contents
Abstract ..................................................................................................................................... ii
Table of Contents ..................................................................................................................... iii
List of Figures .......................................................................................................................... vi
List of Tables .......................................................................................................................... vii
Acknowledgements ................................................................................................................ viii
Introduction ................................................................................................................................8
1. Literature Review.................................................................................................................10
1.1 Learning Processes.........................................................................................................10
1.2 Ovarian Hormones and Learning Processes ..................................................................19
1.2.1 Effects of Ovarian Hormones on Learning Processes: Animal Studies ................ 21
1.2.2 Effects of Ovarian Hormones on Learning Processes: Human Studies ................. 25
1.3 Stress and Learning Processes .......................................................................................32
1.4 Interactive Effect of Ovarian and Stress Hormones on Learning Processes .................38
2. Present Study .......................................................................................................................43
3. Methods................................................................................................................................45
3.1 Participants .....................................................................................................................45
3.2 Study Design ..................................................................................................................47
3.3 Procedure .......................................................................................................................51
3.4 Probabilistic Classification Learning Task ....................................................................53

iv
3.4.1 Weather Prediction Task Analysis ......................................................................... 55
3.5 Surveys ...........................................................................................................................56
3.6 Lexical Task ...................................................................................................................58
3.7 Hormonal Analyses ........................................................................................................59
3.8 Analysis..........................................................................................................................59
4. Results ..................................................................................................................................61
4.1 Sample Size and Distributions .......................................................................................61
4.2 Demographic and Testing-Related Information ............................................................62
4.3 Descriptive Summary of Data ........................................................................................66
4.4 Weather Prediction Task Performance ..........................................................................71
4.5 Psychophysiological Correlates of the Menstrual Cycle Phase, Ovarian Hormones, and
Stress ....................................................................................................................................72
5. Discussion ............................................................................................................................81
5.1 Study Sample .................................................................................................................82
5.2 Effect of Menstrual Cycle Phase and Ovarian Hormones on Learning Processes ........85
5.3 Effect of Stress on Learning Processes ..........................................................................87
5.4 Interactive Effects of Stress and Ovarian Hormones on Learning Processes ................90
5.5 Limitations and Future Directions .................................................................................92
6. Conclusion ...........................................................................................................................94
References ................................................................................................................................96

v
Appendix A: SurveyMonkey Physiological Questionnaire ...................................................111
Appendix B: Demographic Questionnaire .............................................................................112
Appendix C: Menstrual Cycle Questionnaire ........................................................................114

vi
List of Figures
Figure 1. Brain Regions Recruited in Cognitive and Habitual Learning Processes ................18
Figure 2. Task Structure of the Weather Prediction Task ........................................................36
Figure 3. Mean and Standard Deviations of Salivary Estradiol Levels across the Menstrual
Cycle ........................................................................................................................................72
Figure 4. Means and Standard Deviations of Salivary Progesterone Levels across the
Menstrual Cycle .......................................................................................................................73
Figure 5. The Relationship between Salivary Progesterone Concentrations and Strategy
Dominance Scores ...................................................................................................................74
Figure 6. Correlation between Strategy Dominance Scores and MAQ Debilitating Subscale
Scores .......................................................................................................................................75
Figure 7. Correlation between Strategy Dominance Scores and MAQ Anticipation Subscale
Scores .......................................................................................................................................76
Figure 8. Correlation between Strategy Dominance Scores and MAQ Bothersome Subscale
Scores .......................................................................................................................................77
Figure 9. Correlation between Strategy Dominance Scores and BATM Secrecy Subscale
Scores .......................................................................................................................................78
Figure 10. The Relationship between Salivary Estradiol Concentrations and Strategy
Dominance Scores over Varying Levels of Chronic Stress .....................................................80

vii
List of Tables
Table 1. Comparison of Goal-Directed and Stimulus-Driven Learning Processes .................11
Table 2. Terminology Used to Label the Learning Processes .................................................12
Table 3. Timeline of Session 1 and 2 .......................................................................................53
Table 4. Probabilities Used in the Weather Prediction Task ...................................................54
Table 5. Sample Demographic Information.............................................................................63
Table 6. Testing Session Characteristics .................................................................................64
Table 7. Testing Session and Demographic Characteristics over the Menstrual Cycle ..........66
Table 8. Means, Medians, and Ranges of Scores for the Mood and Chronic Stress Surveys:
PSS and PANAS ......................................................................................................................68
Table 9. Means, Medians, and Ranges of Scores for the Cognitive Flexibility and Menstrual
Beliefs and Attitudes Surveys: CCFQ, MAQ, and BATM ......................................................69
Table 10. Missed Attention Checks and Survey Items ............................................................70
Table 11. Number of Participants Categorized as Using Cognitive, Habitual, or
Undifferentiated Learning Processes .......................................................................................71

viii
Acknowledgements
I would like to express my deepest gratitude to my supervisor, Dr. Annie Duchesne,
whose encouragement and guidance were crucial to this project and my growth as a
researcher. I am eternally grateful for your support as my supervisor and mentor. Thank you
for the meaningful and insightful discussions about the project, your dedication to both the
project and my goals, and your patience and advice over the last four years.
I would like to thank Dr. Heath Matheson and Dr. Jacqueline Pettersen for being a
part of my supervisory committee. Thank you for supporting the development of this project
through your valuable feedback and advice.
I would like to thank my friends in the lab, Jihanne, Saleah, Stephanie, and Molly, for
your invaluable support. I would not have been able to complete this project without your
assistance and advice. Jihanne, thank you for being there for me through the most
challenging part of this journey, and uplifting me whenever things were too difficult.
Lastly, I would like to express my love and gratitude for my family, who gave me
strength throughout this project. Thank you for supporting my educational journey, providing
me with guidance and love at every turn, and being there whenever I needed you.

Introduction
As we interact with our environments, we use a combination of processes to learn about
our surroundings effectively. We implement these learning processes whether we are trying to
find a location, predict a particular outcome based on available information, or decide the value
of different options (as reviewed in Wirz et al., 2018). For example, when we navigate an
environment, we can learn to reach a destination by memorizing the necessary turns or

9
integrating different aspects of the environment to create a map (Wirz et al., 2018). These two
systems are respectively known as habitual and cognitive learning processes 1 (as reviewed in
Schwabe & Wolf, 2013). Habitual learning allows the formation of associations between one
stimulus and the corresponding response, while cognitive learning involves making multiple
associations between different stimuli (Schwabe & Wolf, 2013). Notably, while we concurrently
use habitual and cognitive processes to learn about our environment, individual and contextual
variables can alter the relative recruitment of the learning processes. Then, advancing our
understanding of how people learn requires the knowledge of how learning occurs in a specific
context.
Ovarian hormones are examples of such context. Literature from rodent studies
consistently shows that both estradiol and progesterone affect learning processes, specifically in
spatial navigation (reviewed in Brake & Lacasse, 2018). However, the sparse existing research in
humans shows an effect with an inconsistent direction. Taken together, the existing literature
suggests that ovarian hormones regulate learning processes in humans, but the effect may differ
based on other interacting factors (Hussain et al., 2016; Scheuringer & Pletzer, 2017).
Interestingly, acute exposure to stressful situations as well as high levels of chronic stress have
been found to influence the recruitment of learning processes. However, to date, the potential
interactive effects of stress and ovarian hormone fluctuations on the recruitment of learning
processes remain largely unexplored. This proposal will outline the literature on the individual
and interactive effects of ovarian hormones and stress on learning processes and then outline the

1The learning processes that control an organism’s behaviour can be influenced by many endogenous and

exogenous factors, but the organism itself cannot choose which process is implemented (Packard & Goodman,
2013).

10
present study, which aims to explore the interactive effects of stress and ovarian hormones on the
recruitment of learning processes. Thirty-two naturally-cycling people (i.e., those who are not
taking hormonal contraception) completed a task evaluating learning processes during one of
three targeted menstrual cycle phases. We aimed to test participants when both ovarian hormones
are low, when only estradiol is high, and when estradiol and progesterone are high; these phases
were confirmed using salivary ovarian hormone measures. A chronic stress survey and salivary
cortisol hormone measures assessed stress levels.
1. Literature Review
1.1 Learning Processes
How organisms make decisions or control their behaviour is dependent on two distinct
processes: the presence of a cue in the environment or a goal to be attained (as reviewed in
O'Doherty et al., 2017; see Table 1). The former describes stimulus-driven learning, which
initiates in response to a stimulus and engages in learning based on the absence or presence of
reward alone. Similar to habitual learning processes, the behaviour that results from this learning
is rapid, rigid, and uses less cognitive resources. The latter indicates goal-directed learning,
which results in the evaluation of actions and subsequent implementation of behaviours that help
achieve specific outcomes and goals, such as performing a certain action. Unlike stimulus-driven
learning, goal-directed learning (similar to cognitive learning processes) is slower and more
flexible, and uses more cognitive resources (O’Doherty et al., 2017). Together, these two
different learning processes allow for more adaptive learning based on the organism’s situation
and context.
Table 1

11
Comparison of Goal-Directed and Stimulus-Driven Learning Processes

Reason for behaviour
Speed
Adaptability
Use of cognitive resources

Stimulus-Driven

Goal-Directed

An environmental cue

A specific goal

Rapid

Slow

More rigid

More flexible

Low

High

Viewing learning processes through a lens of reinforcement learning provides a
mechanism of how these two learning processes work2. Reinforcement learning uses the reward
expected from a specific action to determine the value of that action (Daw et al., 2005). This
learning ensures that the greater the expected reward is, the faster the action will be learned and
selected over other actions with less rewarding probability (Daw et al., 2005). The stimulusdriven responses (or habitual learning processes) arise from model-free or cued response
reinforcement learning (O'Doherty et al., 2017). Specifically, model-free reinforcement learning
uses reward prediction error, which refers to the neural signal that indicates the degree of
discrepancies in rewards expected and obtained. Learning from these reward prediction errors
creates action values, allowing people to make their next choice based on the reward they
received after completing a particular action (Doody et al., 2022). Here, one would expect that a
greater reward prediction error directly corresponds to increased learning (McDannald et al.,
2012). Further, this type of learning does not consider the overall structure of the environment;
thus, the resultant behaviours are less affected by outcome value changes since past instances of
rewards are used to make choices (O’Doherty et al., 2017). Conversely, goal-directed responses

2 There are various terms that label the two learning processes, aside from the terminology used in reinforcement

learning (Wirz et al., 2018; Table 2). The terminology used in this thesis will include the cognitive and habitual
learning processes, unless the literature discussed uses specific terminology.

12
(or cognitive learning processes) arise from model-based reinforcement learning. This process
involves creating a cognitive map that represents potential actions and the expected probability
by which each action leads to specific outcomes (O’Doherty et al., 2017). Specifically, modelbased learning uses model-based values instead of action values; these values attend to the future
states (i.e., conditions) and rewards instead of directly connecting possible actions with rewards
(Doody et al., 2022). Learning occurs using this system through state prediction errors, which
signal whether the cognitive map created matches the real environment (Doody et al., 2022). The
difference between state and reward prediction errors lies in whether it signals for a divergence
from the predicted state transition (i.e., moving to a specific state after choosing an action) or
from a previously received reward (Gläscher et al., 2010). So, using model-based learning allows
people to detect and respond to recent changes in their environment, making this a flexible
system (O'Doherty et al., 2017). In sum, model-free learning makes connections between the
action taken and the resultant reward, whereas model-based learning uses a map of potential
states and rewards and the probabilities that link these components. Thus, the reinforcement
learning framework shows that stimulus-driven and goal-directed learning is rooted in distinct
but parallel processes, each sensitive to different aspects of one’s context.
Table 2
Terminology Used to Label the Learning Processes
Processes
Domain
Reinforcement Learning
Navigation

Stimulus-Driven

Goal-Directed

Model-free

Model-based

Stimulus-Response or
Response

Spatial or Place

13
Probabilistic Classification

Habitual or Multi-cue

Cognitive or Single-cue

The idea of different processes constituting learning and behaviour comes from the
multiple memory systems framework. This framework suggests that both cognitive and habitual
learning processes exist within different cognitive domains, and each process relies on different
brain substrates (Schwabe & Wolf, 2013). The two systems exist in tandem (O'Doherty et al.,
2017; Packard & Goodman, 2013) and can benefit each other by providing the other system with
information to function more effectively (O'Doherty et al., 2017). For instance, the model-based
system can provide the model-free system with information about action values, indicating
which choice would have resulted in the maximum reward in a retrospective manner (Doody et
al., 2022). Specifically, in one study, researchers asked participants to provide a subjective value
for the stimuli presented in a two-step decision task. Typically, the two-step decision task
involves participants making two choices in each trial. The first decision always requires the
participants to pick between two stimuli, and this decision leads to one of two possible Stage 2
states. The choices from Stage 1 are related to either Stage 2 states with different probabilities.
For example, a certain choice in Stage 1 can be related to State A with a 30% probability (i.e.,
rare transition) and to State B with a 70% probability (i.e., common transition). These states from
Stage 2 present another decision to be made; the choice from this second stage is
probabilistically related to whether or not the participant receives a reward. In this task, making
the same Stage 1 choice based on whether it was rewarded in the previous trial exhibits the
recruitment of the model-free process as the action was chosen based on the relation between the
action and the actual reward. The model-based process involves using the probabilistic
associations between choices, states, and subsequent rewards to make decisions. More

14
specifically, the model-based process would utilize a cognitive map that represents the
transitions between the Stage 1 choices and Stage 2 states, as well as the probability of receiving
a reward after making certain choices in Stage 2. The task included in this study was modified to
include a subjective value decision. The subjective value decision was made in a model-free
manner based on the stimuli’s shape, which was not relevant to the attributes focused on in the
actual two-step decision task. These values determined whether the previous trial’s transition
type (i.e., common or rare) affected the value given to the shape; in other words, the values
reflected whether the model-based understanding of the task influenced the model-free learning.
The findings showed that model-based input from the transition type was integrated into the
model-free learning to determine these subjective values since the shapes of the stimuli given a
reward in a prior trial after a common transition were assigned more value than the shapes of
those rewarded after a rare transition (Doody et al., 2022). As such, integrating input from these
two processes allows us to effectively learn about our surroundings (O'Doherty et al., 2017). The
multiple memory systems framework then illustrates that the moderating factors in one’s context
can influence both how these processes work separately as well as cooperatively.
The importance of how these learning processes work, either separately or together, lies
in how their recruitment influences psychopathology. Biases in the use of learning processes
have been considered etiopathological to several conditions, with a shift towards habitual
learning over goal-directed being considered a transdiagnostic feature of various compulsive
disorders (as reviewed in Voon et al., 2017). An important body of literature suggests that eating
and stress-related disorders, such as obsessive-compulsive disorder (OCD), substance use
disorder, bulimia nervosa, and post-traumatic stress disorder, have all been attributed to
dysfunctional switching between the learning processes (Berner & Marsh, 2014; Wirz et al.,

15
2018). For instance, OCD is characterized by obsessions, which are intrusive thoughts of the
presence of a threat, and compulsions, which are behaviours repeated in order to avoid this threat
(as reviewed in Gillan & Robbins, 2014). These compulsions that people with OCD experience
are proposed to arise from an increased reliance on habitual learning over cognitive learning
(Gillan & Robbins, 2014). Greater use of habitual learning is also observed in substance use
disorders (as reviewed in Goodman & Packard, 2016). Specifically, people who are dependent
on substances, such as alcohol or tobacco, show increased recruitment of habitual processes and
impairment in cognitive learning (Goodman & Packard, 2016). Altered activation in brain
regions related to learning process recruitment and thus abnormal recruitment of these learning
processes fosters an environment for different psychopathologies (Wirz et al., 2018).
Understanding the factors and conditions that modulate learning processes has important
scientific and clinical implications as it can help us further understand these problems and
provide a foundation on which new solutions can be found. Based on the context, the interaction
between the learning processes proposed by the multiple memory systems framework can allow
organisms to adaptively learn about their environments or can be a marker of different
psychopathologies.
The processes postulated by the multiple memory systems are related to activity in
various brain regions (see Figure 1). A recent meta-analysis by Huang and colleagues (2020),
interrogating the brain correlates of reinforcement learning processes, shows that there is both
overlap and distinction in the brain regions underlying each process. Specifically, model-based
(i.e., cognitive) learning correlates with activity in regions implicated in executive functioning,
such as the prefrontal cortex (PFC) and the cingulate cortex. For example, the lateral prefrontal
cortex (LPFC) can represent action values and integrate different characteristics of the expected

16
reward (as reviewed in Dixon & Christoff, 2014). This region also represents the more complex
features of value-based learning, such as assigning value to abstract ideas and rewards and
comparing the action values of the chosen versus other possible actions (Dixon & Christoff,
2014). Using a variant of the two-step decision task, one study found a role of the dorsolateral
PFC (dlPFC) in the initial learning of the task structure and learning about the relationships
between the task stages (Wittkuhn et al., 2018). Further, the orbitofrontal cortex (OFC)
represents information about the potential rewards and uses this information to determine what
behaviours one engages in (as reviewed in McDannald et al., 2012). Lastly, using an adapted
version of the two-step decision task, the anterior cingulate cortex in mice was found to represent
various aspects of the task structure, such as the probability by which a choice leads to a certain
outcome and the prediction one makes regarding the outcome of their choice (Akam et al., 2021).
On the other hand, model-free (i.e., habitual) learning tends to use the brain regions caudate head
and globus pallidus, and both of these regions are proposed to signal reward prediction errors.
Both model-based and model-free learning processes rely on activity in the ventral striatum
(Huang et al., 2020; McDannald et al., 2012), which decides the learning process employed
(Huang et al., 2020). A review by Gahnstrom and Spiers (2020) also implicated the caudate
nucleus (part of the dorsal striatum) as a region that serves both model-free and model-based
processes by depicting the transition structure. While general similarities exist across cognitive
domains, the regions implicated in the different learning processes differ based on the cognitive
task of interest (Patterson & Knowlton, 2018). Specifically, the meta-analysis by Huang and
colleagues (2020) did not find hippocampal activity to be responsible for the learning processes
used during the two-step task. In contrast, the hippocampus does play a role in spatial (i.e.,
cognitive) learning during navigation (Gahnstrom & Spiers, 2020). Of note, OCD is associated

17
with changes in the activity of the caudate and structure of the putamen (part of the dorsal
striatum) and OFC (Gillan & Robbins, 2014), all regions implicated in the recruitment of
learning processes. In sum, there is evidence showing the role of distinct brain regions in the
recruitment of different learning processes, which reveals the targets by which contextual factors
can modulate learning.

18
Figure 1
Brain Regions Recruited in Cognitive and Habitual Learning Processes

Note. Blue represents brain regions that underlie cognitive learning, green represents regions
implicated in habitual learning, and black represents brain regions responsible for both cognitive
and habitual learning. “File:Brain midsagital view.png” by Mike Birkhead is licensed under CC
BY 4.0.
Learning processes undergo modulation through various experimental factors. For
example, as participants train more with the task of interest, they begin to use habitual learning
more compared to cognitive learning (as reviewed in Packard & Goodman, 2013). Habitual
learning processes can also be recruited more often when there are longer periods of time in
between trials, as compared to shorter time intervals which increase cognitive learning. Further,
having more heterogeneity in the environment while navigating a maze task can shift the balance
towards cognitive learning. However, a homogenous environment will create a shift towards
habitual learning as the information in the environment is not sufficient to form a cognitive map
(Packard & Goodman, 2013). While there are numerous moderators that exist in the context of

19
the task, the recruitment of learning processes can also be influenced by endogenous factors,
such as stress and ovarian hormones (Scheuringer & Pletzer, 2017; Wirz et al., 2018).
1.2 Ovarian Hormones and Learning Processes
The investigation of the influence of ovarian hormones on brain and behaviour has
evolved from a controversial past. Previous attempts to understand brain and behaviour
correlates of the menstrual cycle pursued evidence supporting stereotypical differences between
men and women, such as distinctions in intelligence and emotional capacities (Shields et al.,
2018; Sommer, 1992). This attempt was unsuccessful at corroborating these stereotypical
differences and instead revealed a regulatory role of ovarian hormones on cognitive processes
(Beltz & Moser, 2020; Sundstrom-Poromaa, 2018). This evidence comes in the form of both
administration studies (i.e., assessing changes after pharmacologically increasing hormone levels
to a supraphysiological state) and natural phenomena (i.e., assessing changes that accompany
naturally occurring fluctuations in ovarian hormones; Beltz & Moser, 2020; SundstromPoromaa, 2018). One type of natural phenomenon that can help further understand the regulatory
role of ovarian hormones is the menstrual cycle. Understanding how the menstrual cycle creates
variability in brain and behaviour within menstruating people is one way of rejecting the
previous controversial attempts at studying ovarian hormones.
Most people assigned as female at birth have characteristic changes in their levels of
ovarian hormones, estradiol and progesterone, across their lifetimes. Ovarian hormones are the
principal product of the hypothalamic-pituitary-gonadal (HPG) axis and thus control
reproductive functions (as reviewed in Phumsatitpong et al., 2021). The levels of ovarian
hormones undergo predictable fluctuations within people assigned as female at birth, either on a
monthly basis or across major physiological life events. Ovaries begin producing estradiol and

20
progesterone during gonadarche, which is a period of ovary development occurring between ages
10 to 18 (as reviewed in Beltz & Moser, 2020). After puberty, people begin their menstrual
cycle, which is a recurring event characterized by typical changes in these hormones and spans
from 21 to 37 days (Beltz & Moser, 2020). One major physiological change females experience
is pregnancy, during which the levels of estradiol and progesterone increase to
supraphysiological, or above typical physiological, levels and subsequently return to normal
levels after the birth (as reviewed in Haggerty et al., 2003). At around 51 years old, people begin
to experience menopause, which is the end of the monthly menstrual cycles (Beltz & Moser,
2020). During this time, the levels of estradiol and progesterone start to drop (Beltz & Moser,
2020). Importantly, the effects of ovarian hormones rely on various types of receptors. For
example, the effects of estradiol require binding to the estradiol receptors (ER) ERɑ, ERβ, and G
protein-coupled estradiol receptor 1 (as reviewed in Hara et al., 2015). Similarly, the actions of
progesterone are facilitated through progesterone receptors (PR). The three estradiol receptors
are distributed throughout different brain regions, such as the prefrontal cortex, hippocampus,
hypothalamus, and amygdala, while the progesterone receptors are largely found in the
hippocampus (Hara et al., 2015). Given the location of estrogen and progesterone receptors in
the brain, it is feasible that the lifetime changes in ovarian hormones can enact brain and
behavioural changes.
Due to its recurring nature, the menstrual cycle is a natural phenomenon that creates
characteristic monthly variations in ovarian hormones and subsequently brain and behaviour.
The menstrual cycle is a physiological event that ensures fertilization of the female reproductive
cell, the oocyte (as reviewed in Schmalenberger et al., 2021). This cycle contains a follicular and
a luteal phase. The follicular phase spans from when menses begins to when ovulation occurs

21
and encompasses the development and release of the oocyte. The luteal phase starts after
ovulation and ends the day before the next menses. Here, the follicle that releases the oocyte
becomes the corpus luteum. Broadly, estradiol levels remain low during the early and midfollicular phases and spike in the late follicular phase. This increase in estrogen is a result of the
release of follicle-stimulating hormone (FSH), a byproduct of the HPG axis. Estradiol levels then
experience a drop in the early luteal phase and a moderate peak in the mid-luteal phase.
Progesterone remains low throughout most of the cycle, peaking during the mid-luteal phase
(Schmalenberger et al., 2021). In sum, there is great but systematic variation in the levels of
estradiol and progesterone across the life span of menstruating people. Demonstrating the
importance of the menstrual cycle as a window to variation in ovarian hormones, the literature to
follow focuses on looking at how the menstrual cycle influences brain and behaviour,
specifically learning processes.
1.2.1 Effects of Ovarian Hormones on Learning Processes: Animal Studies
The investigation of how ovarian hormones influence brain and behaviour comes from
the larger field of behavioural neuroendocrinology. The majority of advances seen in behavioural
neuroendocrinology are attributable to research done in animal populations (for example, see
Hilz, 2022). Specifically, the mechanisms of hormone action within the human brain can be
derived from rodent research (Hilz, 2022). As such, a large body of rodent research has revealed
the effect of ovarian hormones on learning processes and proposed mechanisms by which this
influence occurs, which lends support for examining the regulatory role of ovarian hormones
within humans.
The role of estradiol in regulating learning processes is particularly well-established in
rodent literature. For example, Korol and colleagues (2004) used an elevated T-maze to

22
investigate how the recruitment of navigational learning processes (i.e., place and response
learning) fluctuates across the rodent estrous cycle 3. During training, the rats navigated through a
T-shaped maze to find a food reward in one of two open arms. In the probe trial, the navigational
learning process was assessed by placing the rat in the opposite arm compared to the training
phase. Response or habitual learning involves using the same direction turn as the training phase,
whereas place or cognitive learning involves using cues from the room to turn towards the arm
that previously contained the food reward. They found that female rats preferentially used place
learning to successfully find the food reward during the proestrus phase, characterized by high
levels of progesterone and estradiol, compared to other estrous phases. Inversely, the rats used
response learning and thus made the same but incorrect turn more during the estrus phase, when
both ovarian hormone levels were low (Korol et al., 2004). Korol and Kolo (2002) conducted a
similar study, looking at the direct manipulation of ovarian hormones instead of natural variation
across the estrous cycle. In this study, ovariectomized4 rats received an injection of either
estradiol (to reach slightly higher levels than those found in proestrus) or a control substance and
completed a navigational task that forced the animal to use either place or response learning.
Similar to the previous study, the findings showed that compared to the rats receiving a control
substance, the rats injected with estradiol completed the place maze more successfully and the
response maze less successfully (Korol & Kolo, 2002). Further, these consistent effects of
estradiol on learning processes are reflected within brain regions associated with human learning

3 The estrous cycle is the equivalent of the human menstrual cycle within rodents (as reviewed in Marcondes et al.,

2002). This cycle spans around four days and includes the following phases in order: estrus, metestrus, diestrus, and
proestrus. Estrus is characterized by decreasing levels of both progesterone and estradiol. During metestrus and
diestrus, estradiol levels begin to increase and progesterone levels reach a moderate peak and then decrease.
Proestrus phase is characterized by a peak in both estradiol and progesterone (Marcondes et al., 2002).
4 Ovariectomized mice are those who have their ovaries removed surgically, and thus cannot produce ovarian
hormones (Korol & Kolo, 2002).

23
processes. As reviewed by Brake and Lacasse (2018), estradiol administration into the rat
hippocampus results in a targeted improvement in place learning. Similarly, the administration of
estradiol to the medial prefrontal cortex, which is a brain region associated with switching
between learning processes, changes the learning process recruitment from response to place
learning. However, estradiol administration to the rat dorsal striatum specifically impaired
response learning. Taken together, rodent literature consistently shows that estradiol influences
specific brain regions to create a place learning bias that is contingent on the dopaminergic
system.
Although less investigated, rodent studies show that variation in circulating levels of
progesterone modulates navigational learning processes (as reviewed in Brake & Lacasse, 2018).
In a recent study, Lacasse and colleagues (2022) administered ovariectomized rats with either
low levels of estradiol (similar to diestrus), high levels of estradiol, or high estradiol levels paired
with progesterone (similar to proestrus). In the estradiol and progesterone condition,
progesterone administration occurred either 15 minutes, one hour, or four hours before testing
the rats to characterize the time course of its effects. The rats then completed the T-shaped maze,
where they used either response or place learning to find a food reward. Similar to the findings
by Quinlan and colleagues (2008), the results showed that low levels of estradiol administration
corresponded with a bias towards response learning and high levels of estradiol corresponded
with a bias towards place learning. However, progesterone administration reversed this bias seen
in high estradiol conditions, specifically starting one hour after administration (Lacasse et al.,
2022). This reversal in learning process recruitment could be due to progesterone changing
activity in the hippocampus, dorsal striatum, and prefrontal cortex, each of which has
progesterone receptors (as reviewed in Gomez-Perales & Brake, 2022; Lacasse et al., 2022). In

24
sum, rodent research shows that progesterone can eliminate the place or cognitive learning bias
observed with high circulating levels of estradiol, indicating that the two ovarian hormones have
opposing effects.
Rodent research also shows an effect of ovarian hormones on learning processes in
outcome devaluation. For example, a study by Schoenberg and colleagues (2022) investigated
the effects of ovarian hormones on goal-directed (i.e., cognitive) and habit learning in outcome
devaluation, a task that involves the elimination of the value of a previously trained response.
Ovariectomized rats received either low estradiol, high estradiol, or high estradiol and
progesterone (to replicate the levels during diestrus or proestrus) when learning to poke a button
to receive a food reward. After this acquisition, half of the sample underwent outcome
devaluation related to the food reward. Following outcome devaluation, the rats were put in front
of the button again to determine whether they pressed the button to receive more of the food
reward. Unlike the findings in spatial navigation, rats given either low or high estradiol engaged
more in goal-directed learning and successfully extinguished the learned behaviour. However,
the rats given high estradiol and progesterone showed a bias towards habitual learning as the rats
did not show extinction of the learned behaviour. The researchers conducted the same study
again using medroxy-progesterone (MPA) instead of progesterone. As this molecule is similar to
progesterone but does not break down into allopregnanolone (a metabolite of progesterone), this
reiteration of the study aimed to rule out effects of the progesterone metabolite. Interestingly, this
experiment showed that MPA did not confer the same effects as progesterone in the previous
experiment, which indicates that the effects of progesterone instead occur through its metabolites
(Schoenberg et al., 2022). Research in rodents shows consistent effects of estradiol,

25
progesterone, or progesterone metabolites on learning processes across different domains,
indicating that ovarian hormones influence adaptive behaviour overall.
1.2.2 Effects of Ovarian Hormones on Learning Processes: Human Studies
The findings from rodent literature warrant an investigation into the potential systematic
changes in learning processes seen with ovarian hormone fluctuations in humans. The translation
from rodent studies to human populations requires a pseudo experimental approach. In the case
of ovarian hormones, this approach involves using natural hormonal variations across the
menstrual cycle to approximate their effects on brain and behaviour. However, this pseudo
experimental approach has been used infrequently in the field of learning processes. So far,
studies looking at learning processes concentrate on either sex differences, with a few studies
investigating the effects of cycle phase. Generally, there is extensive research looking at the
effects of ovarian hormones on learning processes in rodents, but the research that aims to
translate these findings to human populations is still emerging.
Spatial navigation and the related learning processes are well-characterized in human
populations. This research posits that spatial navigation follows two perspectives (allocentric or
egocentric) and two strategies (Euclidean or landmark-based; Hussain et al., 2016). Allocentric
perspective refers to making associations between several aspects of the outside environment and
following externally-based directions (i.e., north, east, south, and west; Hussain et al., 2016;
Scheuringer & Pletzer, 2017). In contrast, the egocentric perspective involves using directions
with respect to one’s body (i.e., right and left) to navigate (Hussain et al., 2016; Scheuringer &
Pletzer, 2017). The egocentric perspective reflects habitual learning as it relies on using
memorized body turns rather than a representation of the space to navigate through the
environment (Brake & Lacasse, 2018). Conversely, the allocentric perspective reflects cognitive

26
learning as it requires a cognitive map of one’s surroundings (Brake & Lacasse, 2018).
Euclidean navigation uses global landmarks and distances and is most often used in conjunction
with allocentric navigation (e.g., making a map of the environment based on the present cues;
Hussain et al., 2016; Scheuringer & Pletzer, 2017). Like the egocentric perspective, landmarkbased navigation involves using specific environmental landmarks parallel to personal directions
of right and left (e.g. specific cues that inform one to turn right; Hussain et al., 2016; Scheuringer
& Pletzer, 2017). This framework has led the investigation of navigational learning processes in
human populations, specifically in the area of sex differences.
The question of whether ovarian hormones play a role in navigational learning processes
comes from sex differences research in spatial navigation. A recent study by Harris and
colleagues (2019) looked at the performance of female and male participants while completing a
computerized spatial navigation task with directions that used one of four possible combinations
of perspectives and strategies. In this task, Euclidean instructions would use distances to describe
the location, whereas landmark-based instructions would describe the location as relative to a
landmark in the environment. Egocentric instructions use language that is personal to the
individuals (i.e., using words like left or right), whereas allocentric instructions use cardinal
directions (i.e., east or west). For example, participants in the Euclidean and allocentric condition
would be told to move a certain number of spaces in a certain cardinal direction (i.e. north). They
found that male participants were typically faster overall, and female participants had faster
reaction times when using an egocentric perspective and landmark-based strategy than an
allocentric and Euclidean combination. Among other explanations for these sex differences, the
authors also proposed the role of the following hormones: estradiol, progesterone, and
testosterone. The results showed that progesterone was positively correlated with the time it took

27
to complete the navigation maze task, but the hormones did not interact with the type of
directions to influence performance (Harris et al., 2019). As this study controlled for hormone
fluctuations by having women participate exclusively during the mid/late-luteal phase of the
menstrual cycle, testing during different phases of the menstrual cycle could reveal a potential
mediating role of hormones on learning processes. Similar to this study, Noachtar and colleagues
(2022) investigated brain activity and connectivity during navigation using the same task as
Harris and colleagues (2019). However, in this version, instead of having unlimited time to reach
a target and measuring response time as a proxy of navigation, the participants had to reach as
many targets as they could within a time limit. Interestingly, the findings showed no sex
differences in the performance while using any combination of the strategies and perspectives,
and this lack of difference was attributed to the adaptations made to the task. Of note, the task
used in both of these studies focuses on how well participants performed when forced to use a
certain learning process rather than allowing them to employ a learning process themselves. This
discrepancy reflects a different concept than what was targeted in the rodent literature. Further,
this type of task looks at performance while completing a certain behaviour instead of learning
by trying to achieve a reward or goal. While there are mixed findings regarding sex differences
in navigational learning processes, this literature indicates that ovarian hormones, specifically
progesterone, may play a role in the sex differences observed.
To date, only two studies investigated the modulation of learning processes by ovarian
hormones through the proxy of the menstrual cycle phase. One such study is by Scheuringer and
Pletzer (2017), who conducted a mixed-design study to look at sex differences in the recruitment
of learning processes and how these processes change during different stages of women's
menstrual cycles. The spatial navigation task was completed as a two-dimensional computer

28
grid, in which participants had to travel from one place to another using a combination of the
strategies and perspectives described above. They found that using the egocentric perspective
(i.e., habitual process) yielded faster reaction times during the luteal phase (characterized by
moderate estradiol and high progesterone) than the follicular phase (characterized by increasing
estradiol levels and low progesterone), and the opposite was true for the allocentric perspective
(i.e., cognitive process). During the luteal phase, accuracy was greater when using the landmark
strategy and lower when using the Euclidean strategy. Lastly, the authors found that as
progesterone levels increased, irrespective of phase, accuracy increased while using the
egocentric perspective and decreased while using the allocentric perspective. Again, this study
focused on strategy efficiency instead of strategy selection as the participants had to follow
predetermined directions to complete the maze (Scheuringer & Pletzer, 2017). Taken together,
this study shows that progesterone plays a key role in learning processes, but these results come
at the cost of focusing on strategy efficiency without allowing participants to recruit their
preferred strategies.
Only one study investigates the link between learning processes and the menstrual cycle
while focusing on strategy selection. Hussain and colleagues (2016) conducted a betweensubjects study that tested navigational learning processes in the early follicular (i.e., low
hormone levels), ovulatory (i.e. high estrogen levels), or mid-to-late luteal phases (i.e. moderate
estrogen and high progesterone levels). The spatial navigation task involved learning to find
items in a virtual three-dimensional maze with eight arms extending out from the middle of a
circular platform, using either specific turns (i.e., response or habitual learning) or landscape
cues (i.e., spatial or cognitive learning). Here, the probe trials involved asking participants to find
items in four of the eight arms once while the surrounding landscape is visible, and then find

29
items in the other four arms while the landscape was hidden. In this case, participants using the
spatial or cognitive strategy would make more mistakes in the second part of the trial by entering
arms they had already gotten items from in the first part since they no longer have cues from the
landscape; the participants using the response or habitual strategy would make less errors as they
depend on body turns instead of the landscape. Unlike the previously described studies,
participants completed the task using their preferred learning strategy, and the researchers
analyzed the strategy used based on the participant’s responses. Contrary to Scheuringer and
Pletzer (2017), the results showed a preference for the spatial strategy during the mid-luteal
phase and the response strategy during the early follicular and ovulatory phases. The authors
suggested that menstrual-cycle-related changes in hippocampal and caudate nucleus activity
could underlie these preferences in learning processes (Hussain et al., 2016). In sum, studies
focusing on the association between the menstrual cycle and navigational learning processes
yield inconsistent findings, perhaps in part due to task differences and conflation between
strategy selection and efficiency.
Recently, the investigation of the effects of the menstrual cycle on the learning processes
has expanded from the spatial navigation paradigms into the reinforcement learning paradigm.
Diekhof and colleagues (2021) conducted a mixed design study looking at how women recruit
model-based and model-free learning strategies during a Markov two-step decision task in their
early follicular (i.e., low hormone levels) and late follicular (i.e., high estrogen levels) phases.
The sample of women was grouped based on their COMT-genotype 5, which is an indirect

5

The COMT-genotype determines the activity level of an enzyme that metabolizes dopamine in the prefrontal
cortex (Diekhof et al., 2021). The possible genotypes can have either the Val or Met allele. The Val allele creates an
enzyme that is more active than the Met allele, resulting in lower levels of baseline dopamine in Val carriers
(Diekhof et al., 2021).

30
indicator of the activity of the enzyme that degrades dopamine (DA) in the prefrontal cortex.
Specifically, the Met genotype, which metabolizes DA more slowly, is associated with greater
levels of dopamine in the PFC, while the Val genotype will result in less dopamine. The Markov
two-step decision task was modified to include two stages: the drift stage (i.e., when the
probabilities associating actions with rewards change across the trials) and stable stage (i.e.,
when these probabilities remain constant). The study found that model-free learning (i.e., using
previous rewards to guide actions) was used more during the late follicular phase compared to
the early follicular phase, while there were no differences in the model-based scores (i.e., guiding
actions based on a map of the probabilities by which different actions are related to specific
outcomes); the effect was specific to the drift stage. However, this effect as well as overall
decreased model-based scores was specifically found in Met carriers (in individuals who have
higher DA levels in the PFC); the Val carriers showed no change based on cycle phase. Thus, it
is possible that the increase in estradiol could have increased the use of model-free learning, and
this effect was specific to people with higher dopamine availability in the prefrontal cortex. Of
note, this study did not include the mid-luteal phase to allow the approximation of progesterone
effects on learning processes. The authors also proposed some systems that could be implicated
in the differential recruitment of learning processes. It is possible that estradiol inhibits activity
in the connections between frontostriatal regions (i.e., the prefrontal cortex and striatum) and the
activity of COMT in the PFC, both of which underlie model-based learning. Further, estradiol
could also increase dopamine transmission in the dorsolateral striatum, creating a shift towards
model-free learning (Diekhof et al., 2021). These results are consistent with the findings from
Hussain and colleagues (2016), reinforcing that a habitual learning process is used more in
higher estradiol conditions. Taken together, human studies are aligned with rodent literature in

31
that dopamine plays a role in how estradiol influences learning processes, but not in the direction
of estradiol’s effect.
In addition to investigating learning processes, some work has explored the effect of the
menstrual cycle on different underlying components of model-based and model-free learning.
For example, Veselic and colleagues (2021) used a reinforcement learning task to look at how
the administration of supraphysiological levels of estradiol influenced reinforcement learning
within men. During this task, the participants were required to choose between two stimuli, each
related to a reward by specific probabilities. However, these probabilities continually changed
throughout the task. The results showed that men who received an estradiol dose showed greater
reward sensitivity and faster learning compared to those who received a placebo (Veselic et al.,
2021). Similarly, Bayer and colleagues (2020) focused on activity within the nucleus accumbens
(a part of the ventral striatum) that was specific to prediction errors when naturally cycling
women completed a similar task as Veselic and colleagues (2021). The researchers found that
when naturally cycling women were injected with varying doses of estradiol, ranging from
physiological to supraphysiological concentrations, the activity in the nucleus accumbens that
signals reward prediction errors showed a positive correlation with estradiol concentrations.
However, this change in activity did not manifest as change in behaviour and learning during the
reversal learning task (Bayer et al., 2020). Taken together, these studies show differing effects of
exogenous estradiol on the reward learning between men and women. Focusing on another
aspect of learning, a study by Lewis and colleagues (2022) looked at how different value-based
decision-making aspects changed between people taking oral contraceptives (thus having low
levels of endogenous ovarian hormones), naturally cycling women in the early follicular phase,
and naturally cycling women in the periovulatory phase. Specifically, one of the tasks that the

32
participants completed was the delayed discounting task, which looked at whether the
participants chose more immediate but small rewards or future larger rewards, reflecting habitual
and goal-directed processes respectively. The findings showed that there was no difference
between the groups in terms of delayed discounting (Lewis et al., 2022). Thus, research so far
shows various effects of estradiol on different aspects of reward learning, indicating that
estradiol has an involved and complex role in cognitive versus habitual learning.
The current literature reveals a complex picture of how ovarian hormones affect learning
processes in humans. Studies focusing on navigational learning processes yield mixed results,
likely due to task differences and varying attention to strategy selection versus efficiency
(Hussain et al., 2016; Scheuringer & Pletzer, 2017). When reinforcement learning processes are
studied, findings corroborate one navigational learning study to reinforce that estradiol increases
recruitment of habitual learning (Diekhof et al., 2021). Interestingly, this is in direct contrast to
what the majority of rodent literature shows (Brake & Lacasse, 2018; Lacasse et al., 2022).
Furthermore, human studies looking at the effects of estradiol on different facets of learning
processes reveal a complex role of this hormone (Bayer et al., 2020; Lewis et al., 202; Veselic et
al., 2020). However, the effects of progesterone are understudied and unclear. Thus, while there
is an effect of ovarian hormones on learning processes in humans, focusing on these two
hormones alone cannot provide a comprehensive understanding.
1.3 Stress and Learning Processes
The importance of stress across all cognitive domains is substantiated by research on its
cognitive ramifications. Stress is a fundamental psychophysiological process that results from
threats in the environment and has various dimensions that are strong predictors of different
cognitive functions, particularly learning (as reviewed in Joels et al., 2006). Unlike ovarian

33
hormones, there is a substantial body of literature showing consistent effects of both acute and
chronic stress on learning processes. These studies generally show a shift from goal-directed to
habitual learning after exposure to stress (as reviewed in Wirz et al., 2018). The interconnections
of the reproductive and stress systems (as reviewed in Phumsatitpong et al., 2021) provide
various potential pathways by which stress modulates the effects of ovarian hormones on
learning processes.
Stress responses create various changes within the physiology of the body, and thus the
brain. During this response, individuals release two types of hormones that can influence
neurons: noradrenaline and corticosterone (as reviewed in Joëls et al., 2006). Noradrenaline
comes from a specialized region within the brain, the locus coeruleus, and has a rapid, short-term
effect on neurons. Corticosteroids come from the adrenal cortex and have a slow but long-term
effect on the neurons exposed; this hormone is cortisol when released in humans. Cortisol release
is a product of the actions of the hypothalamic-pituitary-adrenal (HPA). In this process, the
stressor leads to the release of corticotrophin-releasing hormone (CRH), resulting in the
formation and release of the adrenocorticotropin hormone (ACTH). ACTH then acts on the
adrenal gland to begin corticosterone production. The effects of cortisol are exerted on the brain
through either the mineralocorticoid (MR) and glucocorticoid receptors (GR). While GR are
located throughout the brain, MR are found in brain regions such as the prefrontal cortex and
hippocampus. As such, one major change from stress that has been widely researched is its
effects on learning and memory (Joëls et al., 2006). Exposure to stressors thus creates
physiological and biochemical changes, which ultimately influence learning.
The effects of stress on learning strategies are visible across different domains. Similar to
ovarian hormones, one cognitive domain that stress can influence is navigational learning. In the

34
body of literature focusing on stress, navigational learning processes include spatial learning
(i.e., cognitive learning) that allows us to learn how several cues in the environment relate to one
another in order to create a "cognitive map" (Epstein et al., 2017) and stimulus-response or S-R
learning (i.e., habitual learning) that creates associations between one cue and the corresponding
response to navigate (Wirz et al., 2018). For example, in a study by Schwabe and colleagues
(2007), participants first underwent a social stressor or control condition, and then were
presented with small three-dimensional replica of a room, which included a table with four cards
on the surface, as well as other items that acted as spatial cues. Over the course of 12 trials,
participants were asked to select one card that they thought would have the word “win” hidden
on the bottom. During each of these trials, the replica was spun to provide the participants with a
new view of the space, but the cards stayed in the same position. On the 13th trial, a proximal
cue (i.e., a plant placed next to the win-card), was moved to a different location. In this case,
participants used S-R learning if they selected the card closest to the relocated proximal cue, and
they used spatial learning if they chose the correct card using a cognitive map of the entire space.
Here, the researchers found that the presence of stress increases the use of S-R learning
compared to spatial learning (Schwabe et al., 2007). Furthermore, stress can impact the relative
use of cognitive and habit processes in probabilistic classification learning, which involves
learning the probabilities by which different cues are associated with certain outcomes (Wirz et
al., 2018). Probabilistic classification learning can be assessed using the weather prediction task
(see Figure 2; Gluck et al., 2002). During this task, participants are shown one to three cards out
of four possible cards with different stimuli, and asked to decide whether the cards predict rain or
sun. The shapes are correlated with each outcome by a specific probability, and these
probabilities are combined when the shapes show up together in a trial (Gluck et al, 2002).

35
Cognitive learning involves using a model of the probabilities and focusing on single cues within
entire patterns to predict the outcome, whereas habit learning involves connecting the entire
pattern to a certain outcome (Wirz et al., 2018). Specifically, using the entire pattern involves
non-declarative processes (i.e., making predictions with rules that cannot be conveyed with
words), whereas using single cards to make predictions requires declarative knowledge (i.e.,
being able to convey the rules used to make predictions; Gluck et al., 2002; Schwabe & Wolf,
2013). Schwabe and colleagues (2013) looked at acutely stressed individuals' performances on a
probabilistic classification task after taking either an MR antagonist or a placebo. They found
that stress resulted in the balance between the learning processes to shift towards habit learning.
In addition, the participants who received MR antagonists did not experience this shift, which
means that MR action underlies the observed shift after stress (Schwabe et al., 2013). Despite the
cognitive domain of interest, exposure to acute stress shifts the balance between learning
processes towards habitual learning.

36
Figure 2
Task Structure of the Weather Prediction Task

Note. The participant first sees the fixation cross on the screen. This screen is followed by the
presentation of the pattern of cues and rain and sun options. After the participant makes their
choice or runs out of time, they see feedback in the form of a happy or sad face.
Another cognitive domain in which stress has exerted effects in terms of learning strategy
recruitment is instrumental learning. For example, a study by van Ruitenbeek and colleagues
(2021) looked at how goal-directed and habitual learning changes with acute physical and social
stress. The stress and control conditions were split based on whether the participants were given
methylphenidate (which results in an increase in dopamine and noradrenaline) or placebo,
creating four groups in total. The participants then completed an instrumental learning task
where they chose between two options based on the stimulus presented on the screen in order to
receive one of two virtual food rewards. After, the participants were asked to eat one of the food
rewards until satiated, which led to outcome devaluation. Within an MRI, participants were then
asked to complete the task again, this time instructed to give the wrong answer to the food

37
reward that was recently devalued; this action assessed goal-directed learning. The findings
showed that the acute stress created a bias towards habitual learning, reflected by the participants
still using the prior learned response for the food reward that underwent outcome devaluation.
The administration of methylphenidate did not affect the behaviour seen in either the stress or
control groups. The stress condition showed altered activity in some brain regions, such as the
putamen, amygdala, inferior and middle frontal gyrus, and insula, compared to the control
condition. The changes in brain activity in some of these regions underwent reversal in the
condition that was also administered methylphenidate (van Ruitenbeek et al., 2021). This
particular finding shows a parallel between ovarian hormones and stress in that the effects of
both on learning processes are modulated by the dopaminergic system. In sum, the presence of
stress again shifts behaviours towards habitual processes in instrumental learning, but the effect
on the underlying brain regions interacts with dopaminergic and noradrenergic systems.
Chronic stress can influence learning processes in a manner similar to that seen in acute
stress. For example, a study by Radenbach and colleagues (2015) used a within-subjects design
to test whether interindividual differences in chronic stress may influence their use of modelbased and model-free processes during a Markov two-step task. The male participants attended
two sessions, one that employed a social stressor and another that employed a control
manipulation. Interestingly, the results showed no shift towards using model-free processes when
faced with an acute social stressor. However, participants who experienced more stressful life
events and thus had greater chronic stress showed an impairment of the model-based processes
after experiencing the acute stressor (Radenbach et al., 2015). This study reveals an interaction
effect of chronic and acute stress on learning processes. Similarly, a study by Lenow and
colleagues (2017) employed the patch foraging task to investigate how both acute and chronic

38
stress influence decision making processes. The patch foraging task involves harvesting the
maximum number of apples by choosing between foraging from the same patch of apples (i.e.,
exploitative or habitual process) or exploring other patches (i.e., explorative or cognitive
process). The results showed that both an acute stressor and measures of chronic stress were
associated with greater use of the exploitative strategy (Lenow et al., 2017). Thus, chronic stress
is an important modulating variable that can affect how acute stress and perhaps other factors
influence learning processes.
Stressors shift the balance towards increased use of habitual learning through the action
of the resultant hormones on different brain regions. Corticosterones and noradrenaline interact
with the amygdala, and the connections this region has with the dorsal striatum and hippocampus
can result in a shift toward habitual learning (Wirz et al., 2018). Acute and chronic stress may
decrease activity in the prefrontal cortex through the action of corticosterones, which reduces the
recruitment of cognitive learning (as reviewed in Harms, 2017). Chronic exposure to elevated
levels of corticosterones can result in dendritic retraction in the prefrontal cortex, which further
reduces the activity in that brain region. The activity in the striatum is generally unchanged;
however, the decreased activity in the prefrontal cortex and hippocampus may lead to a reliance
on the striatum, thus increasing the recruitment of habitual learning (Harms, 2017). Seeing how
these brain regions parallel the regions affected by ovarian hormones, both stress and ovarian
hormones can interact to affect learning processes.
1.4 Interactive Effect of Ovarian and Stress Hormones on Learning Processes
Most studies look at the effects of ovarian hormones and stress on learning processes
separately, ignoring potential interactions between these factors. The gonadal and stress axes can
intertwine at different levels. For example, there is physiological evidence of interactions among

39
components of the HPA and HPG axes (as reviewed in Phumsatitpong et al., 2021). Indeed, a
study by Barel and colleagues (2018) showed that the changes in the HPA axis and sympathetic
nervous system (SNS) reactivity after a psychological stressor can be accounted for by levels of
testosterone, estradiol, and progesterone. Thus, these physiological interactions between both of
hormones could influence learning processes differently compared to the effects of stress and
ovarian hormones alone. There are some commonalities of the brain regions influenced by stress
and ovarian hormones, such as the hippocampus, prefrontal cortex, and dorsal striatum (Hussain
et al., 2016; Wirz et al., 2018). The effects of stress on learning processes may differ based on
the menstrual cycle phase, and the effects of the cycle phase on learning processes may differ
based on the levels of acute and chronic stress.
There have been only two studies that focus on this interaction between stress and
ovarian hormones in rodent samples. Notably, ter Horst, Kentrop, de Kloet, and Oitzl (2013)
investigated the interaction effect of stress and the estrous cycle phase on rodent spatial learning
using the circular hole board paradigm. This board contains either open or closed holes along its
circumference, leading to the mice’s home cages. During training, the mice had to explore and
find a single open hole to reach their home cages; this hole was marked by a water bottle next to
it. Before the probe trial, the researchers moved the water bottle to another newly opened hole
opposite the first opened hole. The mice recruited spatial learning if they used cues around the
room and entered the previously opened hole to reach their home cages, and recruited S-R
learning if they entered the newly opened hole using the water-bottle cue. Half of the mice
underwent restraint to induce stress after completing the first training trial, while the other half
did not. The authors found that control mice used spatial and S-R learning equally, whereas more
of the stressed mice used spatial learning compared to the S-R learning. This effect was driven

40
by the differences seen in the estrus phase (i.e., low estradiol and progesterone). While the
stressed and control mice in the proestrus and diestrus phases did not differ in learning process
recruitment, the stressed mice in the estrus phase used the spatial strategy more than the control.
Since the estrus phase has low hormone levels and the other phases have either high estradiol or
progesterone levels, the decreased levels of ovarian hormones in the estrus phase may play a part
in how the mice responded to stress (ter Horst et al., 2013a). Another similar study by ter Horst,
Kentrop, Arp, Hubens, de Kloet, and Oitzl (2013) looked at how the deletion of the MR gene in
the forebrain influences navigational learning in either acutely stressed or non-stressed mice
using the circular hole board. Inconsistent with the previous study, the authors could not find a
main effect of stress and estrous phase on learning processes across all mice (ter Horst et al.,
2013b). Emerging rodent literature looking at how stress and reproductive axes interact reveals
inconsistent findings, highlighting the need to understand this interaction within human
populations.
One study inadvertently uncovered a potential interaction between the reproductive and
stress axes while studying the effects of stress on learning processes. Specifically, Schwabe and
colleagues (2009) found that high levels of glucocorticoids correlated with reduced S-R learning
and increased spatial learning in female participants. Among other explanations for these
findings, the authors suggested that moderately high stress can shift the balance towards S-R
learning, but further increases in stress can restore the balance between the two learning
processes and result in poor performance in learning. The authors only tested women using oral
contraceptives in their sample; this criterion may be an underlying cause of the unexpected
results (Schwabe et al., 2009). The presence of oral contraceptives and the resultant changes in
the hormonal milieu may reverse the effects of stress on learning processes.

41
To date, there has been one study that focused on how stress and the menstrual cycle
phase interact to influence navigational learning processes in human participants. In a betweensubjects study, McHale (2019) tested naturally cycling women with a computerized Hex Maze
under stress or control conditions, during either the early follicular, ovulatory, or mid-luteal
phases. The Hex Maze consists of arms protruding from a central platform, each marked by gray
spheres like the radial arm maze. During the training trials, participants learned to look for a
hidden platform in one of the arms, which was marked by a coloured sphere. During the test
trials, the coloured sphere was switched with a gray sphere from a different arm. The possible
learning processes included the allocentric strategy (i.e., using the extra-maze cues to navigate),
the egocentric-cue strategy (i.e., using the coloured sphere), and the egocentric-response strategy
(i.e., turning the same way as before to reach the platform). The results showed that participants
did not differ in the recruitment of learning processes across the menstrual cycle phases and
stress conditions, and there was no interaction effect between the two factors. Among other
reasons, the author attributed the lack of effect of the menstrual cycle phase to the task being
used, since this task allowed participants to recruit more than two types of learning processes
while navigating (McHale, 2019). While this study does not reveal an interaction effect of stress
and menstrual cycle phase within navigational learning, it is possible that interaction effects
appear when the investigation is expanded into the domain of probabilistic classification and
reinforcement learning.
Aside from the physiological interaction of stress and reproductive axes, the menstrual
cycle can also interact with stress through its biopsychosocial characterization. Here, in addition
to the biological dimension of the cycle, the menstrual cycle can be understood as a sociocultural
phenomenon. Gendered conceptualizations of the menstrual cycle tend to consider menstruation

42
and other psychophysiological manifestations related to the cycle as negative, creating an
environment where menstruation is a topic that must be kept a secret and that makes a person
more unclean and inferior compared to others (Johnston-Robledo & Chrisler, 2013). These
beliefs surrounding menstruation make people feel self-conscious and shameful, while also
directing what they can or cannot do (Johnston-Robledo & Chrisler, 2013). The negative beliefs
transform phenomena such as menstruation into a chronic source of stress that can vary in
severity depending on the degree of adherence to such beliefs (Marván et al., 2014). Aside from
the stress associated with menstruation, awareness of the stigma related to menstruation may also
create changes in how people respond to stressors. For example, Doyle and Molix (2018)
revealed that the experience and expectation of stigma can influence one’s reactions to social
stressors, suggesting that the presence of menstrual cycle-related stigma can influence how stress
is experienced. As such, the social construction of the menstrual cycle can become a stressor, and
this stress itself can potentially interact with the menstrual cycle to influence learning processes.
The interplay of stress and ovarian hormones on learning processes has been overlooked
throughout most of the related literature. For example, studies that look at the effects of ovarian
hormones and the menstrual cycle phase on learning processes do not include measures of
cortisol. Conversely, studies looking at the effects of stress on learning processes tend to either
exclude female participants or test them during a single menstrual cycle phase and not measure
ovarian hormone levels. It is possible that studies focusing on the effect of either stress or
ovarian hormones on learning processes miss the interactive effect of both factors, creating the
inconsistencies we see in human literature about ovarian hormones and learning processes.
Further, most of the studies investigating the effect of ovarian hormones rely on the menstrual
cycle, but do not consider how menstrual beliefs and attitudes may contribute to variation in

43
learning processes. To gain a clearer understanding of how ovarian hormones and stress
influence learning processes, studies should investigate their interaction while adopting a
biopsychosocial conceptualization of the menstrual cycle.
2. Present Study
People use either flexible (i.e., cognitive learning) or rigid (i.e., habitual learning)
processes to learn about their environment (Wirz et al., 2018). There are various factors that can
bias which learning process is employed (Packard & Goodman, 2013). One such factor is stress,
which creates a preference for habitual learning over cognitive learning (Wirz et al., 2018). Other
influencing factors include ovarian hormones, estradiol and progesterone. The consensus from
rodent research is that estradiol favours cognitive learning and progesterone reverses these
effects of estradiol (Brake & Lacasse, 2018; Lacasse et al., 2022). However, human research has
conjured an unclear picture of how estradiol and progesterone affect learning processes. The
inconsistencies in this body of literature could arise from task differences and differing focuses
on strategy efficiency versus selection. It could also be due to the lack of focus on the interactive
effect of stress and ovarian hormones. Most studies focus on either the effects of stress or
ovarian hormones, not attending to the interactive effects of the stress and reproductive axes.
Moreover, these studies do not consider the role of the social dimension of the menstrual cycle,
which can act as a source of stress. Considering that variation in stress and ovarian hormones
may interact to regulate learning processes, the present thesis aims to:
1. Determine whether the menstrual cycle phase or circulating levels of estradiol and
progesterone influence learning processes used during a probabilistic
classification learning task.
2. Determine whether perceived chronic stress influences learning processes.

44
3. Explore whether perceived chronic stress moderates the association between
learning processes and ovarian hormone levels or menstrual cycle phase.
Naturally cycling participants attended two testing sessions. They completed a weather
prediction task during one session and a different cognitive task in the other. We expected
greater recruitment of cognitive learning during the late follicular phase and habitual learning in
the mid-luteal phase. Estradiol would positively correlate with the recruitment of cognitive
learning and progesterone levels would positively correlate with the recruitment of habitual
learning. Further, perceived chronic stress would correlate positively with habitual learning. We
also expected an interaction effect of perceived chronic stress and ovarian hormones on learning
processes. Specifically, estradiol levels would attenuate and progesterone levels would
emphasize the effect of chronic stress. Importantly, this study would contribute to the growing
body of research investigating how ovarian hormones influence learning processes across the
menstrual cycle and would elaborate on the interaction between reproductive and stress axes.
Broadly, this work would add to the knowledge of how the use of learning processes can be
contingent on the interactions between various contextual factors.

45
3. Methods
3.1 Participants
A sample of 32 students at the University of Northern British Columbia (UNBC) was
recruited using convenience sampling. Recruitment was done using two distinct approaches: the
SONA system and on-campus recruitment. The UNBC Psychology Research Participation
SONA system was used to recruit undergraduate students in the UNBC Psychology Department
during the January 2023 semester (from January to April). In the Spring 2023 semester (from
May to June), on-campus recruitment was conducted in addition to the SONA system. The oncampus recruitment was accomplished by advertising the study throughout the campus.
Participants were given either bonus psychology course credits or monetary compensation for
their participation (3 credits or $30 for each session). This project received ethics approval from
the University of Northern British Columbia (UNBC) Research Ethics Board (reference number
E2019.1209.074.03(a)).
A developing body of research shows that priming menstrual- or gender-related
stereotypes can influence how participants perform during cognitive tasks or react to stress
(Doyle & Molix, 2018; Romans et al., 2012; Wister et al., 2013). Therefore, to reduce this bias in
the participants, the recruitment protocol was designed to minimize the awareness participants
had about the menstrual cycle during the study. When they signed up, participants were not
informed about the true objectives of the project and were instead told that the study was an
investigation into the practice effects on the recruitment of learning processes. They received
information about the true objectives during the debriefing period at the end of the study and
were asked to not share this information with other students.

46
Upon registering for the study, participants completed a survey that included questions to
confirm their eligibility. Participants included people who were assigned as female at birth and
were naturally cycling (i.e., have not taken any hormonal medications and contraceptives within
three months before the study). There were a number of exclusionary criteria implemented in this
project. Firstly, participants were excluded if they were diagnosed with a reproductive disorder
(such as endometriosis, polycystic ovarian syndrome, premenstrual dysphoric disorder, and
ovarian cysts). Reproductive disorders or their treatments could influence various aspects of the
HPG axis, ultimately creating distinct hormonal profiles compared to naturally cycling women
(Knudsen et al., 2004; De Leo et al., 2016; Schmidt et al., 2017; Vannuccini et al., 2022).
Participants were also excluded based on the usage of chronic medications within the past three
months (including hormonal medications and contraceptives) and a history of hormone-related
surgeries (i.e., oophorectomy or hysterectomy). Hormone-related surgeries and the use of
hormonal medications could alter the typical variation of ovarian hormones during the menstrual
cycle (Hampson, 2020; Sarrel et al., 2016). Lastly, people with a history of brain injuries and
psychopathology were excluded to avoid any biases in the recruitment of learning processes (see
Berner & Marsh, 2014; Voon et al., 2017; Wirz et al., 2018). In sum, each of these exclusionary
criteria ensured that there were minimal outliers in the participants’ hormonal profiles or use of
learning processes.
Premenstrual syndrome (PMS) and premenstrual dysphoric disorder (PMDD) are two
affective conditions that typically arise before menstruation (Takeda, 2023). Evidence regarding
the hormonal basis of these conditions is yet emerging, but it is clear that PMS and PMDD can
considerably change mood and cognition (Castro et al., 2021; Le et al., 2020; Takeda, 2023). As
such, researchers tend to exclude participants diagnosed with PMS and PMDD from cognitive

47
studies. However, there are various socio-structural considerations in addition to these affective
and cognitive dimensions of PMS and PMDD. Here, the experience of PMS and PMDD can be
related to individual experiences of stress and family life as well as to the broader social
construction of premenstrual symptoms (Coughlin, 1990; Ussher, 2003; Ussher et al., 2007).
Considering the biosocial nature of PMS and PMDD, we used a screening tool to explore the
occurrence of moderate to severe PMS or PMDD within our sample but did not exclude data
based on this categorization.
3.2 Study Design
A between-subjects study was conducted to test how the menstrual cycle phase, hormone
levels, and chronic stress levels independently and interactively influenced the use of learning
processes in a probabilistic classification learning task. This study was included in a larger
project investigating various correlates of the menstrual cycle and ovarian hormone levels. As
such, participants attended two sessions during two of three possible menstrual cycle phases and
completed a probabilistic classification learning task during one of the two sessions. Initially, the
session in which this task was done was fully randomized. In order to prioritize data collection
for the probabilistic classification learning task due to a limited timeline, the full randomization
protocol was changed to 2-to-1 task assignment in the Spring 2023 semester. As participants
signed up, this method involved assigning two participants to complete the probabilistic
classification learning task in the first session, one participant to complete the other cognitive
task during the first session, and so on. During the Spring 2023 semester, participants were also
given the option to attend either one or two sessions. Participants who completed one session
only were assigned the probabilistic classification learning task.

48
Participants’ sessions were booked according to specific phases of the menstrual cycle. In
the Winter 2023 semester, the two sessions took place in two of the following three phases: early
follicular (with low estrogen and progesterone), late follicular (with high estrogen and low
progesterone), and mid-luteal (with moderate estrogen and high progesterone). A series of six
menstrual cycle phase pairs were generated from these three phases and participants were
randomized to be tested using one of the six combinations. The timing of the sessions was
determined by implicitly collecting menstrual cycle data from the participants. This data
collection was done by sending participants emails every other day containing a link to a
physiological questionnaire on Survey Monkey that asked about their menstrual status (see
Appendix A). The questionnaire also included distractor questions (inspired by the Daily Life
Questionnaire by Schwartz et al., 2012) to prevent the participants from discovering the true
objective of the study. Once menstruation was reported, a 28-day template of the cycle was used
to schedule the sessions during the assigned phases. This method of scheduling study sessions
was rigorous but also demanding for the participant. On average, participants completed the
physiological survey for 47.38 days, and some participants had to complete the physiological
survey for multiple months during their participation (range = 14 - 91 days). Additionally, there
was considerable attrition (N = 21), often due to participants not answering the physiological
surveys or booking the study sessions. Considering the methodological challenges resulting from
this initial approach, modifications were made to the session scheduling procedures.
Starting in the Spring 2023 semester, participants were asked to book their study sessions
right after registering for the study. Participants could choose between completing one or two
sessions that took place 11 to 12 days apart, based on the participants’ availability. Spacing the
sessions 11 to 12 days apart could accommodate two menstrual cycle phases within the typical

49
menstrual cycle lengths. These changes in the study session scheduling procedures created a
system where participants’ menstrual cycle phases were determined retrospectively. Similar to
the previous session scheduling process, participants received the physiological surveys every
other day starting from when they signed up to when they finished their second session. The
information from these surveys was used to retrospectively determine the menstrual cycle phases
in which participants completed their two sessions. To determine which menstrual cycle phases
participants attended their study sessions, information from both prospective and retrospective
methods was used. The use of retrospective and prospective data minimized the recollection bias
in reporting menstruation (Jukic et al., 2008). Thus, in addition to the physiological surveys,
participants also completed a Menstrual Cycle questionnaire during the second session, which
asked them to indicate the start date of their last menses and the average length of their menstrual
cycle. Upon completion of their second session, participants tested in the Winter 2023 semester
were asked to provide the date of the following menses through an online questionnaire.
Participants tested in the Spring 2023 semester were asked to complete four more physiological
surveys after their second session, each four days apart, since half of the participants from the
Winter 2023 semester did not remember to report their following menses.
Considering that there was variation in how the menstrual cycle information was
collected, the number of physiological surveys answered, and the reliability of the different
pieces of information, a score representing confidence in the menstrual cycle phase estimation
was also computed. This score was based on two metrics: quality of cycle length information and
menses dates. The quality of the menses date was given one to five points, based on the time
between when it was reported and when the period actually occurred; this score reflected how
prospectively or retrospectively the date was provided. The quality of the cycle length

50
determination was given one to four points based on the reliability of the information used. The
cycle length could have been derived using a combination of the following information, from
most to least reliable: prospective menses dates, retrospective menses dates, or a 28-day
template. The scores of these two factors were summed; higher scores on both metrics were
assigned to participants who provided more prospective information, and the score decreased the
more retrospective or low quality information was being used. If no information about the
menstrual cycle was available (i.e., participants did not answer the physiological questionnaires
regularly or did not complete the Menstrual Cycle questionnaire), then participants’ phases were
categorized as “undefined.”
In addition to influencing the participants’ performance, the awareness of the menstrual
cycle can influence how the researcher would behave during the study sessions. For example, a
study by Roberts and colleagues (2002) showed that people who displayed a sign of
menstruation were seen as less agreeable and competent by others. As such, the experimenters
were kept blind to the menstrual cycle phase and menses date of the participants they were
responsible for testing to avoid any potential experimenter bias or research-related
misconceptions. For the first phase of recruitment, there were three experimenters who were
given the role of both tracker and tester. As a tracker, the experimenters were assigned to track
certain participants’ date of menses and book their experiment sessions based on the menstrual
cycle phases they were assigned to. Once the participant reported starting their menses, the
tracker assigned the participant to be tested by another experimenter (i.e., the tester). Thus, the
tester was completely blind to the participant’s cycle phase and date of menses. In the second
phase of recruitment, there were two experimenters, and each had the role of tracker and tester.
The tracker was responsible for booking the experiment sessions and tracking when the

51
participant started their menses. Thus, the tester was blind to the participants’ menses date. These
methods of blinding were used whenever possible. However, there were instances when other
experimenters were not available for the testing, in which case the tracker had to test the
participant they were tracking.
3.3 Procedure
On average, the first session spanned two hours and the second session spanned 2.5
hours. In the first session, the participants started by filling out a consent form and demographic
questionnaire. Then, the participants completed the first set of saliva samples, including a passive
drool sample for the ovarian hormone measurement and a salivette sample for the cortisol
measurement. After these samples, the participant completed a mood questionnaire. Next, the
participants received instructions about the probabilistic classification learning task or
probability learning task (not focused on in this thesis), after which they were asked to complete
the task. After an optional short break, the participants completed either a spatial working
memory task or interoceptive task (either of which are not focused on in this thesis). They then
completed another cortisol salivette sample and mood questionnaire. Participants ended the
session by completing a set of questionnaires related to chronic stress and cognitive flexibility.
The second session followed the same steps as the first session. However, the participants
did not complete a demographic questionnaire during this session. They were also asked to
complete a lexical task after the second saliva sampling. At the end of the second session, the
participants were debriefed about the study's purpose and asked about their prior knowledge of
the study’s purpose and their level of comfort in understanding the task instructions or
questionnaires. After being debriefed, participants completed the questionnaires focused on
menstrual beliefs and attitudes, premenstrual symptoms, and menstrual cycle characteristics.

52
They were also asked for their consent to use their data once they learned about the actual
objectives of the study. Participants who signed up for one session only followed the steps of the
first session, with the addition of the debriefing session, menstrual beliefs and attitudes
questionnaires, premenstrual symptoms questionnaire, menstrual cycle characteristics
questionnaire, and post-debriefing consent form towards the end of the session. The timeline of
the two sessions is displayed in Table 3.

53
Table 3
Timeline of Session 1 and 2
Session 1

Session 2

Introduction and consent form

10 minutes

10 minutes

Demographic Questionnaire

5 minutes

N/A

Saliva Sampling 1

30 minutes

30 minutes

PANAS 1

2 minutes

2 minutes

Weather Prediction or Probability
Learning Task

20 minutes

20 minutes

Rest (optional)

5 minutes

5 minutes

Spatial Working Memory or
Interoceptive Task

20 minutes

20 minutes

Saliva Sampling 2

5 minutes

5 minutes

PANAS 2

2 minutes

2 minutes

Questionnaires 1

15 minutes

15 minutes

LexTALE

N/A

5 minutes

Debriefing

N/A

5 minutes

Questionnaires 2

N/A

20 minutes

Estimated Total Time

2 hours

2.5 hours

Note. Each participant completed the Weather Prediction Task in session 1 or 2. The other
session included a different cognitive task (not focused on in this thesis).
3.4 Probabilistic Classification Learning Task
The weather prediction task (WPT) was adapted from Gluck and colleagues (2002) to
investigate cognitive and habitual learning in probabilistic classification learning. This task was
administered on Gorilla Experiment Builder, a web platform specifically used to conduct
psychological research. The WPT requires participants to look at up to three cards with distinct

54
shapes and decide whether the weather pattern signalled by the combination of these cards is rain
or sun (see Figure 2). After making this choice, they are shown feedback in the form of either a
happy or sad face. Each card, displaying different shapes, is associated with each weather pattern
by different probabilities (see Table 4). The probabilities associated with each card are combined
when a pattern of multiple cards appears. Participants completed 100 trials, split into four blocks.
There was a one-minute break between each block (Gluck et al., 2002; Wirz et al., 2018).
Table 4
Probabilities Used in the Weather Prediction Task
Shape

Probability of sun as outcome

Square

.756

Diamonds

.575

Circle

.425

Triangle

.244

Note. These probabilities are adapted from Gluck and colleagues (2002) and Zerbes (2021).
Participants could solve this task using either a multi-cue, one-cue, or a singleton strategy
(Gluck et al., 2002). The multi-cue strategy reflects habitual learning, where participants
consider all the cards present to determine the weather pattern; here, the choice made after seeing
the entire pattern resembles a stimulus-response action (Gluck et al., 2002; Wirz et al., 2018).
The singleton and one-cue strategies reflect the use of cognitive learning, with the participants
parsing through the different cues to predict the weather outcome. With the singleton strategy,
the participants respond to patterns containing one card based on the learned probabilities and
guess on trials with patterns containing multiple cards. The one-cue strategy involves

55
determining the weather outcome indicated by a pattern using the presence or absence of one
particular card (Gluck et al., 2002; Wirz et al., 2018).
3.4.1 Weather Prediction Task Analysis
The analysis of weather prediction task learning strategies consisted of comparing the
participants’ responses for each trial to ideal response profiles corresponding to each strategy
(Gluck et al., 2002; Zerbes et al., 2020). A best-fit score for each profile determined which
strategy the participant's responses best resemble, calculated using (1). Here, model M refers to
the learning strategy for which the fit score is being calculated, P is the specific pattern of cards,
#sun_expectedP,M is the number of times it is expected for participants to respond to pattern P
with “sun” if they are using strategy M, #sun_actualP is the number of times the participant
actually responded to pattern P with “sun”, and #presentations P is the number of times pattern P
shows up in the task. This score was calculated for the multi-cue, singleton, and four possible
one-cue strategies. The participant was categorized as using the strategy with the lowest best-fit
score. Based on previous studies, 0.15 was considered the tolerance level of the best-fit scores; if
all of the best-fit scores were greater than 0.15, the participant was categorized as using an
undefined strategy (Gluck et al., 2002; Zerbes et al., 2020).
=

∑ (#

_
∑ (#

,

−#

_
)

)

(1)

Further, a strategy dominance score was calculated by subtracting the multi-cue best-fit
score from lowest of the single-cue scores (i.e., singleton or one-cue scores). While the best-fit
score categorically showed which strategy participants were using, the strategy dominance score
numerically indicated the degree of preference for one strategy over the other. In this case,
positive scores meant there was a bias toward the multi-cue strategy and negative scores showed

56
a bias toward the single-cue strategies. The strategy dominance score was calculated for
participants who were categorized or undefined learners (Zerbes et al., 2020).
Aside from strategy analysis, accuracy was also calculated for the task. Responses were
considered correct if they had a greater probabilistic association with the card pattern shown in
the trial (Gluck et al., 2002; Zerbes et al., 2020). For example, if a pattern of cards was related to
the outcome of rain with a probability of greater than 0.5, then the correct answer for that pattern
would be rain. Patterns related to both outcomes with probability of 0.5 were excluded from the
accuracy analysis (Gluck et al., 2002; Zerbes et al., 2020).
3.5 Surveys
The surveys were administered on Gorilla Experiment Builder. Each survey, except for
the demographic questionnaire, included an instructed-response attention check (see Kung et al.,
2018). These attention checks matched the wording of the rest of the items in the survey, with
the addition of an instruction that details how the participant should respond. For example, the
attention check for one of the surveys stated “In the past month, how often did you feel unable to
cope? Please answer “very often” or “four” for this question.” The attention check questions
were used to determine which participants were not focused during the questionnaire. Before
analyzing the survey data, items were reverse scored as needed and attention checks items were
examined for failed responses. If participants missed an item in a survey or failed the attention
check, their data for that survey was excluded from the analysis.
In the first session, each participant completed the demographic questionnaire, a 13-item
questionnaire that provided a description of the sample (see Appendix B). The questions asked
about the participants’ sex and gender, age, university program, ethnicity, handedness,
medication use, and first language.

57
Participants also completed state surveys (mood and chronic stress) during both sessions
and trait surveys (menstrual attitudes and beliefs and cognitive flexibility) either during the first
or second session.
Mood was assessed using the Positive and Negative Affect Scale (PANAS; Watson et al.,
1988). The PANAS is a 20-item questionnaire that includes a number of positive and negative
emotions, and participants are asked to indicate the level of each positive and negative feeling
they are currently experiencing. This questionnaire provided descriptive insight into the positive
and negative dimensions of the participants’ mood. The PANAS was done before and after the
cognitive tasks to assess the mood at those time points as well as find if there is any difference in
the mood over the course of the tasks.
Chronic stress was measured using the Perceived Stress Scale (PSS; Cohen et al., 1983).
The PSS is a 14-item questionnaire that measured participants’ level of chronic stress by asking
questions about the level of stress they perceive from their recent life events. This questionnaire
was done towards the end of each session.
Cognitive flexibility was assessed using the 20-item Cognitive Control and Flexibility
Questionnaire (CCFQ; Gabrys et al., 2018). The questionnaire had two subscales: “Cognitive
control over emotion” (which included items related to attention and inhibition) and “Appraisal
and coping flexibility” (which included items related to coming up with various coping strategies
and appraising situations with various approaches). Overall, this questionnaire provided
measures of the participants’ perceptions of their levels of cognitive flexibility, which is a
concept linked to learning processes (Wirz et al., 2018).
Menstrual-related beliefs and attitudes were assessed using two questionnaires: the
Menstrual Attitude Questionnaire (MAQ; Brooks-Gunn & Ruble, 1980) and the Beliefs About

58
and Attitudes Toward Menstruation (BATM; Marván et al., 2006) questionnaire. The MAQ is a
33-item questionnaire that includes subscales that describe menses as being debilitating,
bothersome, natural, predictable, and having no effect. The BATM is a 45-item questionnaire
that includes the following subscales: secrecy (i.e., items about hiding one’s menstrual status),
annoyance (i.e., items about labelling menstruation as annoying), proscriptions and prescriptions
(i.e., items listing the things menstruating people should or should not do), disability (i.e., items
describing menstruation as an obstacle in their daily activities), and pleasant (i.e., items
describing menstruation as a sign of well-being). Data from these surveys allowed the
determination of whether any social aspects related to the cycle phases can influence learning
process use.
Participants also completed the Premenstrual Symptom Screening Tool (PSST; Steiner et
al., 2003) and Menstrual Cycle questionnaire (see Appendix C) at the end of the second session.
The 19-item PSST allowed insight into whether the participants experienced premenstrual
dysphoric disorder (PMDD) or moderate to severe premenstrual syndrome (PMS). The
Menstrual Cycle questionnaire asked about the regularity of the participants’ menstrual cycles,
the date when their last period started, and the average length of their menstrual cycle. This
information was used to derive or confirm the phases in which the participants were tested.
3.6 Lexical Task
With the expansion of recruitment from the psychology department to students across
UNBC, a task called the Lexical Test for Advanced Learners of English (LexTALE) was added
to the study to gain an understanding of the levels of English proficiency within the sample
(Lemhöfer & Broersma, 2012). The task was administered through Gorilla Experiment Builder,
and participants were shown 60 words on the screen, with 40 being real words and 20 being

59
nonwords. The participants had to decide if the presented word was a real word or not. Scoring
the task involved individually determining the percent of words and nonwords the participants
correctly classified and finding the average of these two percentages. Higher scores indicated a
higher level of English proficiency. The scores could also categorize participants into either
lower or upper intermediate or advanced language levels based on the Common European
Framework, a system used to assess and categorize language skills (Lemhöfer & Broersma,
2012).
3.7 Hormonal Analyses
After the collection, the passive drool and salivette samples were placed in a -20C
freezer. The passive drool samples were subsequently transferred to a -80C freezer. The saliva
samples were then thawed and analyzed using enzyme-linked immunosorbent assay (ELISA).
This immunoassay for the passive drool samples was done using standardized Salimetrics
Salivary Estradiol Enzyme Immunoassay Kit (Salimetrics Assay #1-3702) and Salimetrics
Progesterone Enzyme Immunoassay Kit (Salimetrics Assay #1-1502). The immunoassay for the
salivette samples was done using Salimetrics Salivary Cortisol ELISA Kit (Salimetrics Assay #13002). The assays were carried out according to the protocols outlined by Salimetrics.
3.8 Analysis
The demographic information and testing characteristics (i.e., testing time, testing day,
session order, and time between sessions) were summarized for the entire sample as well as
according to the assigned menstrual cycle phase.
Normal distribution of the survey data, hormone levels, and strategy dominance scores
were assessed by visually inspecting frequency histograms. Outliers were evaluated based on a
three standard deviation threshold. Survey scores for participants who failed the attention check

60
or missed survey items were excluded from the analysis. To explore whether the current sample
is comparable to similar populations (e.g., student populations), means, ranges, and medians
were computed for each variable and compared to the descriptive statistics of samples from other
studies.
One-way ANOVA was conducted to investigate mean differences across menstrual cycle
phases with respect to salivary hormones, learning strategy, chronic stress levels, mood, and
menstrual attitude and beliefs. Additionally, since the menstrual cycle phase estimation was done
using different types of information with varying levels of confidence, we also assessed possible
mean differences in the confidence scores of the cycle phase estimation based on the assigned
menstrual cycle phase. The individual and interactive effects of circulating estradiol and
progesterone on the strategy dominance scores were computed using multiple regression
analysis. Significant effects were decomposed using pairwise comparisons. The effect of stress
on learning processes was investigated by conducting a multiple regression analysis to
investigate if both PSS scores and salivary cortisol levels predicted strategy dominance scores 6.
Since negative mood can be indicative of acute stress (Watson et al., 1988), correlational
analysis was conducted between the PANAS negative subscale scores (baseline and after
completing the task) and strategy dominance scores. The possible interacting effects between
ovarian hormones and stress on learning processes were assessed by conducting a multiple
regression analysis looking at how PSS scores, cortisol levels, and ovarian hormone levels
together influence strategy dominance scores. Finally, in order to explore how attitudes and

6 While cortisol saliva samples were acquired during the study, the constrained timeline did not allow for the assay

to be completed prior to the submission of the initial thesis. Instead, a Pearson correlation was conducted between
PSS and strategy dominance scores. The analysis including salivary cortisol will be presented at the thesis defense.

61
beliefs about the menstrual cycle may correspond to variation in stress and learning processes, a
correlational matrix exploring correlations between learning processes, chronic stress, cortisol
levels, mood, and menstrual attitudes and beliefs was also computed. Further, Pearson
correlations were conducted to investigate if strategy dominance scores are related to BATM and
MAQ survey subscales. The statistical analysis was done using R, SPSS, and Excel.
4. Results
4.1 Sample Size and Distributions
The weather prediction task was completed by 32 participants. Menstrual cycle phase
estimation revealed that the assigned phases spanned the entire menstrual cycle. Therefore, the
menstrual cycle phases were grouped into early follicular (n = 10), follicular (including mid and
late follicular phases; n = 11), and luteal phases (including early, mid, and late luteal phases; n =
7). Participants who were tested during ovulation or did not provide enough information for the
phase estimation (n = 4) were excluded from the ANOVA, leaving a sample size of 28 for this
analysis. There were also four contaminated passive drool saliva samples, which were excluded
from the analyses with hormonal levels.
The majority of the data from the variables were normally distributed (i.e., strategy
dominance scores; “cognitive control over emotions” subscale of the CCFQ; positive affect
subscale of the PANAS; debilitating, bothersome, anticipation, and natural subscales of the
MAQ; all subscales of the BATM; and salivary estradiol levels). The distributions of the denial
subscale of the MAQ, negative affect subscale of the PANAS, and salivary progesterone levels
were positively skewed. There was a trend for an outlier in each of the following distributions:
PSS, BATM prescriptions and proscriptions subscale, and CCFQ appraisal and coping flexibility
subscale. There was one outlier in the distributions of the PANAS negative affect subscale at

62
times 1 and 2. All analyses including the PANAS negative affect subscale excluded the data
from this participant.
4.2 Demographic and Testing-Related Information
Most participants identified as female, were 17 to 25 years old and were either Caucasian
or South Asian (see Table 5). Since the ethnicity was self-reported, participants reported either
their ethnicity or nationality; the answers were included regardless of the report. The majority of
participants spoke English as their first language and were right-handed. Most participants were
in their first year of post-secondary education and were enrolled in the psychology program.
Eleven participants were categorized as having moderate to severe PMS and five participants
were categorized as having PMDD. The Weather Prediction Task was done mostly during the
first session (see Table 6). Most sessions including the WPT took place during the weekdays.
The time of day during which these sessions took place was evenly distributed between 10 am,
1:30 pm, and 5 pm. The average number of days between Sessions 1 and 2 was 13.6 (range = 7 48).

63
Table 5
Sample Demographic Information
n (%)
Gender

n (%)
Self-Reported Ethnicity/Nationality

Female

31 (96.9)

Caucasian

10 (31.3)

Non-binary

1 (3.1)

South Asian

9 (28.1)

East Asian

2 (6.3)

Age Range
17-25

26 (81.3)

Nigerian

2 (6.3)

25-30

3 (9.4)

Middle Eastern

1 (3.1)

30-35

2 (6.3)

Black

1 (3.1)

35+

1 (3.1)

Ugandan

1 (3.1)

Asian

1 (3.1)

English as First Language
Yes

21 (65.6)

Program of Study

No

10 (31.3)

Psychology

20 (62.5)

Health Sciences

3 (9.4)

Handedness
Right

29 (90.6)

Nursing

2 (6.3)

Left

2 (6.3)

Computer Sciences

2 (6.3)

3 (9.4)

Biochemistry

1 (3.1)

Finance and Economics

1 (3.1)

Medication Use
Year of Study
First

16 (50.0)

General Studies

1 (3.1)

Second

3 (9.4)

English

1 (3.1)

Third

5 (15.6)

Anthropology

1 (3.1)

Fourth

8 (25.0)

Premenstrual Symptoms
PMS

11 (37.9)

PMDD

5 (17.2)

64
Table 6
Testing Session Characteristics
n (%)
Session for WPT
Session 1

22 (68.8)

Session 2

10 (31.3)

Time of Day
10 am to 1 pm

9 (28.1)

1:30 pm to 4:30 pm

10 (31.3)

5 pm to 8 pm

13 (40.6)

Day of the Week
Weekday

25 (78.1)

Weekend

7 (21.9)

Seventeen participants completed the LexTALE, and the LexTALE data of one
participant was excluded as they answered yes to all words presented. The mean score of this
task was 80.94% (range = 51.25 - 100%). As most participants scored greater than 60%, the
majority of the sample was placed in the upper intermediate or advanced language level in the
Common European Framework (Lemhöfer & Broersma, 2012). One participant was placed in
the lower intermediate language level. Upon inspection of their data, the participant was retained
in the analysis as their scores were not considered outliers.
The timing of the sessions and day of the week (i.e., weekend vs weekday) was evenly
distributed across the menstrual cycle phases (see Table 7). The mean confidence score of phase
estimation was 6.10 (SD = 2.42, median = 6.50, range = 2.00 - 9.00). There was no difference in

65
the confidence scores of cycle estimation between the different menstrual cycle phases (p = .94).
There was an imbalance in terms of how many participants completed the WPT in session 1
versus 2, based on the menstrual cycle phase. Specifically, more people who completed the WPT
in the first session were in their early follicular phase. The majority of the people doing the WPT
in the second session were in their follicular phase. The number of participants who were
categorized as having PMDD or moderate to severe PMS were evenly distributed across the
menstrual cycle phases. There were fewer participants who spoke English as their first language
in the luteal phase compared to the early follicular and follicular phases.

66
Table 7
Testing Session and Demographic Characteristics over the Menstrual Cycle
Early Follicular

Follicular

Luteal

Ovulation

Session 1

8

5

5

3

Session 2

2

6

2

0

10 am to 1 pm

2

5

2

0

1:30 pm to 4:30 pm

3

2

2

3

5 pm to 8 pm

5

4

3

0

Weekday

7

8

6

3

Weekend

3

3

1

0

1.41 (0.33)

1.61 (0.21)

1.72 (0.43)

Session for WPT

Time of Day

Day of the Week

Salivary Hormones
Estradiol (pg/mL)
Progesterone (pg/mL)

267.58 (141.77)

236.95 (132.34) 300.85 (117.53)

English as 1st Language

70%

82%

43%

33%

LexTALE Scores (%)

70.00

90.00

74.38

76.25

PMS

2

4

3

1

PMDD

1

2

1

1

Premenstrual Symptoms

Note. This table does not include the participant tested during an unassigned phase.
4.3 Descriptive Summary of Data
The mean, median, and range of scores on the PSS, PANAS, CCFQ, BATM, and MAQ
are summarized in Tables 8 and 9. Further, the number of participants who failed the attention

67
check or missed survey items are listed in Table 10. The MAQ subscale means from this sample
were comparable with the sample in the original article introducing this survey (Brooks-Gunn &
Ruble, 1980). The BATM subscale means from this sample match the means seen in American
and Mexican samples from previous studies (Marván et al., 2005; Marván et al., 2006). The
CCFQ subscales means for this sample and the sample from the original article (Gabrys et al.,
2018) are similar. This sample’s PANAS scores for time 1 and 2 match the scores seen in the
original article (Watson et al., 1988). Lastly, the PSS scores for this sample were slightly higher
than the sample in the original article (Cohen et al., 1983).

68
Table 8
Means, Medians, and Ranges of Scores for the Mood and Chronic Stress Surveys: PSS and
PANAS
Mean (SD)

Median

Range

Positive Time 1

30.86 (7.64)

31.50

17.00 - 44.00

Positive Time 2

28.89 (7.86)

29.00

12.00 - 41.00

Positive Difference Score

-2.19 (3.68)

-2.00

-10.00 - 4.00

Negative Time 1

15.03 (5.16)

14.00

10.00 - 32.00

Negative Time 2

14.37 (5.49)

13.00

10.00 - 34.00

Negative Difference Score

-0.41 (3.64)

0.00

-9.00 - 9.00

29.93 (7.79)

30.50

17.00 - 51.00

PANAS

PSS

69
Table 9
Means, Medians, and Ranges of Scores for the Cognitive Flexibility and Menstrual Beliefs and
Attitudes Surveys: CCFQ, MAQ, and BATM
Mean (SD)

Median

Range

Cognitive Control over Emotion

3.80 (1.23)

3.83

1.44 - 6.00

Appraisal and Coping Flexibility

5.04 (0.87)

5.22

2.56 - 6.44

Debilitating

4.20 (1.11)

4.21

1.25 - 6.42

Bothersome

4.65 (0.80)

4.50

3.33 - 6.33

Natural

5.07 (1.20)

5.50

1.75 - 7.00

Anticipation

4.98 (1.15)

5.25

1.75 - 7.00

Denial

2.58 (1.36)

2.14

1.00 - 6.43

Secrecy

1.63 (0.56)

1.50

1.00 - 3.58

Annoyance

3.56 (0.73)

3.54

2.17 - 4.67

Prescription

2.39 (0.80)

2.44

1.00 - 4.56

Disability

2.90 (0.91)

2.60

1.40 - 5.00

Pleasant

3.01 (0.83)

3.00

1.50 - 4.83

CCFQ

MAQ

BATM

70
Table 10
Missed Attention Checks and Survey Items
Failed Attention Check

Missing Data

Positive Time 1

3 (10.3%)

3 (9.4%)

Positive Time 2

3 (10.3%)

3 (9.4%)

Negative Time 1

3 (10.3%)

3 (9.4%)

Negative Time 2

3 (10.3%)

3 (9.4%)

PSS

2 (6.9%)

3 (9.4%)

MAQ

2 (7.4%)

5 (15.6%)

PANAS

Debilitating

4 (12.5%)

Bothersome

5 (15.6%)

Natural

3 (9.4%)

Anticipation

3 (9.4%)

Denial

3 (9.4%)

BATM

5 (17.2%)

3 (9.4%)

Secrecy

3 (9.4%)

Annoyance

6 (18.8%)

Prescriptions/Proscriptions

4 (12.5%)

Disability

3 (9.4%)

Pleasant

3 (9.4%)

CCFQ

5 (17.9%)

4 (12.5%)

Cognitive control over emotion

2 (6.3%)

Appraisal and Coping flexibility

0

71
The ranges of concentration of ovarian hormones detected in saliva were within the
detection range of the immunoassay (see Table 7). The salivary estradiol concentrations
corresponded to those expected across the menstrual cycle (Gandara et al., 2007). The salivary
progesterone concentrations were higher than what has been documented for the same population
(Delfs et al., 1994).
4.4 Weather Prediction Task Performance
The mean accuracy of the participants during the Weather Prediction Task was 62.64%
(SD = 11.43, range = 39.77 - 93.18%, median = 61.93%), which is similar to what was seen in a
previous study (Schwabe & Wolf, 2012). The mean strategy dominance score was -0.05 (SD =
0.04, median = -0.05, range = -0.129 - 0.064). Most participants were categorized as using
cognitive strategies (i.e., singleton or one-cue strategies; see Table 11). This bias towards using
cognitive strategies is similar to what is seen in non-stressed samples within previous studies
(i.e., Schwabe & Wolf, 2012; Zerbes et al., 2020). The number of participants that used
undifferentiated strategies in this sample is also similar to what is seen in previous studies (i.e.,
Schwabe & Wolf, 2012).
Table 11
Number of Participants Categorized as Using Cognitive, Habitual, or Undifferentiated Learning
Processes
Strategy

n (%)

Cognitive

23 (71.9)

Habitual

3 (9.4)

Undifferentiated

6 (18.8)

72
4.5 Psychophysiological Correlates of the Menstrual Cycle Phase, Ovarian Hormones, and
Stress
There were no significant differences in levels of ovarian hormones (see Figures 3 and 4),
chronic stress levels, positive or negative mood, and menstrual attitudes and beliefs across the
different menstrual cycle phases (all p > .05). Based on visual exploration of the hormonal data,
the mean hormone levels per cycle phase do not follow the typical patterns expected with the
menstrual cycle.
Figure 3
Mean and Standard Deviations of Salivary Estradiol Levels across the Menstrual Cycle

Note. Four participants were excluded from this figure due to blood contamination in the saliva
samples and three more participants were excluded for being tested during ovulation or an
unassigned phase. The sample size for this figure included 25 participants (early follicular phase:
n = 8; follicular phase: n = 10; luteal phase: n = 7).

73
Figure 4
Means and Standard Deviations of Salivary Progesterone Levels across the Menstrual Cycle

Note. Four participants were excluded from this figure due to blood contamination in the saliva
samples and three more participants were excluded for being tested during ovulation or an
unassigned phase. The sample size for this figure included 25 participants (early follicular phase:
n = 8; follicular phase: n = 10; luteal phase: n = 7).
There was no main effect of the menstrual cycle phase on strategy dominance scores (p >
.602). When investigating whether salivary levels of estradiol and progesterone predicted the
variation in learning processes, a trend for a negative association between circulating levels of
progesterone and strategy dominance score was observed (β = -.392, p = .055, n = 28).
Specifically, there was a greater cognitive bias with higher levels of progesterone (see Figure 5).
The regression model showed no association between strategy dominance scores and estradiol (p
> .05).

74
Figure 5
The Relationship between Salivary Progesterone Concentrations and Strategy Dominance
Scores
0.1

Strategy Dominance Scores

0.05

0
0

100

200

300

400

500

600

-0.05

-0.1

-0.15

Salivary Progesterone Concentrations (pg/mL)

Note. Four participants were excluded from this figure due to blood contamination in the saliva
samples, leaving a sample size of 28 participants. This figure shows the relationship between
salivary progesterone concentrations and strategy dominance scores but does not display the
regression model.
No significant association between the strategy dominance score and the PSS score was
observed (r(26) = -.10, p = .602). In the initial correlational analysis, significant correlations
between the strategy dominance score and two MAQ subscales were observed (see Figure 6 and
7). Specifically, there were significant negative correlations between strategy dominance scores
and the debilitating and anticipation subscales (all p < .05). However, upon inspection, the
correlations were seen to be driven by one data point. Another correlational analysis was then

75
done after removing that specific data point. The previous correlations between strategy
dominance scores and the MAQ debilitating and anticipation subscales became non-significant
(all p > .05). Instead, there was a significant positive correlation between the MAQ bothersome
subscale and strategy dominance scores (r(22) = .46, p = .023; see Figure 8). Here, participants
showing a more habitual bias in learning also had stronger beliefs about menstruation being a
bothersome event.
Figure 6
Correlation between Strategy Dominance Scores and MAQ Debilitating Subscale Scores
0.1

Strategy Dominance Scores

0.05

0
1

2

3

4

5

6

7

-0.05

-0.1

-0.15

MAQ Debilitating Subscale Scores
Note. Four participants were excluded as they did not complete part or all of the MAQ, and two
participants were excluded as they failed the attention check. The sample size for this figure
included 26 participants.

76
Figure 7
Correlation between Strategy Dominance Scores and MAQ Anticipation Subscale Scores
0.1

Strategy Dominance Scores

0.05

0
1

2

3

4

5

6

7

8

-0.05

-0.1

-0.15

MAQ Anticipation Subscale Scores

Note. Three participants were excluded as they did not complete part or all of the MAQ, and two
participants were excluded as they failed the attention check. The sample size for this figure
included 27 participants.

77
Figure 8
Correlation between Strategy Dominance Scores and MAQ Bothersome Subscale Scores
0.04

Strategy Dominance Scores

0.02

0
-0.02

2

3

4

5

6

7

-0.04

-0.06
-0.08
-0.1

-0.12
-0.14

MAQ Bothersome Subscale Scores

Note. Five participants were excluded as they did not complete part or all of the MAQ, two
participants were excluded as they failed the attention check, and one participant was excluded
as they were a multivariate outlier. The sample size for this figure included 24 participants.
There was also a significant positive correlation between strategy dominance scores and
the secrecy subscale on the BATM (r(24) = .68, p < .001; see Figure 9). In other words,
participants with a more cognitive bias had less belief in keeping menstruation a secret. This
correlation remained significant with the removal of the multivariate outlier (r(23) = .50, p =
.011). Finally, there was a significant negative correlation between the strategy dominance scores
and the appraisal and coping flexibility subscale of the CCFQ (r(25) = -.51, p = .006). However,
this correlation was lost once the multivariate outlier was removed (p = .159).
Figure 9

78
Correlation between Strategy Dominance Scores and BATM Secrecy Subscale Scores
0.1

Strategy Dominance Scores

0.05
0
0

0.5

1

1.5

2

2.5

3

3.5

4

-0.05
-0.1

-0.15
BATM Secrecy Subscale Scores
Note. Three participants were excluded as they did not complete part or all of the MAQ, and
three participants were excluded as they failed the attention check. The sample size for this
figure included 26 participants.
A regression analysis was conducted to investigate how variation in ovarian hormones
and varying levels of chronic stress predicts strategy dominance scores. Due to the limited
sample size, separate regression analyses were conducted to investigate the interacting effects of
estradiol and progesterone with chronic stress. Results from the linear regression analysis
investigating the interactive effect of salivary estradiol concentrations and PSS scores in
predicting variation in strategy dominance scores revealed a significant interaction between
estradiol and chronic stress levels on predicting strategy dominance scores (β = .499, p = .021, n
= 24). The interaction reveals that the association between salivary estradiol concentrations and
strategy dominance scores is influenced by the levels of reported chronic stress (see Figure 10).
Specifically, people scoring lower than one standard deviation below the mean showed a

79
negative correlation between salivary estradiol concentrations and strategy dominance scores.
For people scoring within one standard deviation of the mean, there was no correlation between
salivary estradiol and strategy dominance scores. Finally, people scoring higher than one
standard deviation above the mean showed a positive correlation between salivary estradiol
concentration and strategy dominance scores. In other words, participants who had high levels of
reported chronic stress showed a habitual bias with higher levels of salivary estradiol, whereas
participants who had low levels of stress showed a cognitive bias with higher levels of salivary
estradiol.

80
Figure 10
The Relationship between Salivary Estradiol Concentrations and Strategy Dominance Scores
over Varying Levels of Chronic Stress

0.04
0.02

Strategy Dominance Scores

0
0.5

1

1.5

2

2.5

-0.02

-0.04
-0.06

-0.08
-0.1

Low Stress
Mid Stress

-0.12

High Stress

-0.14
Salivary Estradiol Concentrations (pg/mL)
Note. This figure shows the simple regressions of the strategy dominance scores and salivary
estradiol levels in each level of stress. Low Stress include cases with PSS scores that are lower
than one standard below the mean. Mid Stress include cases with PSS scores that are within one
standard deviation from the mean. High Stress include cases with PSS scores that are higher than
one standard deviation above the mean. Four participants were excluded from this figure due to
blood contamination in the saliva samples, two participants were excluded as they did not
complete part or all of the PSS, and two participants were excluded as they failed the attention
check. The sample size for this figure included 24 participants.

81
5. Discussion
Research investigating the relative recruitment of learning processes has identified
ovarian hormones and stress as modulating factors, but this literature is accompanied by various
issues. For example, there is limited work focusing on the effects of ovarian hormones, and the
results have contradicted the consistent findings from rodent literature. On the other hand,
research consistently shows that chronic and acute stress can create a habitual bias in learning
(Wirz et al., 2018), but these studies often tend to exclude female participants or constrain
ovarian hormone fluctuations (i.e., Radenbach et al., 2015; Wirz et al., 2017a, 2017b; Zerbes et
al., 2020). Additionally, there is ample evidence of the interactions between the HPG and HPA
axes (Phumsatitpong et al., 2021) and there is support for the menstrual cycle being a more
biopsychosocial phenomenon rather than strictly biological (Johnston-Robledo & Chrisler,
2013). However, there is limited research that conceptualizes the menstrual cycle as a
biopsychosocial phenomenon or investigates how the interaction between the HPG and HPA
axes can influence learning processes. As such, this study aimed to replicate the effect of stress
on learning processes in a probabilistic classification learning task within a sample of naturallycycling menstruating people. Further, the study investigated how salivary estradiol and
progesterone or the social dimension of the menstrual cycle can influence the recruitment of
learning processes. Lastly, the study examined if stress and ovarian hormones interact to
influence the way learning processes are recruited. Our preliminary findings suggest that in a
sample of naturally cycling people, conventional measures of chronic stress do not correspond
with changes in learning processes. However, various negative menstrual attitudes and beliefs
create a bias towards habitual learning. Further, a trend for an association between progesterone
levels and learning processes was observed. Finally, a significant interactive effect of estradiol

82
and chronic stress on learning processes was observed. Taken together, these preliminary
findings emphasize the importance of both investigating the interactive effect of ovarian
hormones and chronic stress and viewing the menstrual cycle as a biopsychosocial phenomenon
in such studies.
5.1 Study Sample
The sample for this study was demographically diverse, despite being small. Further,
there were no biases in the testing characteristics, with the exception of session order; here, more
participants completed the WPT in session 1 than in session 2. This bias was expected as
obtaining WPT data was prioritized to accommodate the Master’s project timeline. There were
no differences in strategy dominance scores, chronic stress, and mood across these two sessions,
suggesting that this session effect did not influence the data. This study also included a lexical
task to understand if there were any differences in the variables based on language differences.
The results showed that over 90% of the sample had at least an upper intermediate understanding
of English, which means there were no notable language barriers in understanding task
instructions or questionnaire items. Further, no outlying patterns in the data were observed from
the participant who scored below the upper intermediate level, suggesting that the difference in
verbal fluency did not bias the results. We also observed a difference in the phases in terms of
people using English as a first language. However, this difference was not seen in verbal fluency
levels, which suggests that the difference in the participants’ first language likely did not
influence the results. As such, the language differences and session effects present in this study
did not create differences in the results.
One issue arising with this sample was that half of the participants were categorized as
having moderate to severe PMS or PMDD. Compared to community samples, the proportion of

83
people with PMS in this study was somewhat similar (38% in the current study, 20-30% in
community samples), but there was a considerably higher number of people categorized as
having PMDD (17% in the current study, 1.2-6.4% in community samples; Steiner et al., 2003;
Takeda et al., 2022). PMS is characterized by the experience of physical or mood-related
symptoms that affect one or more domains of a person’s life. PMDD is a more severe version of
this disorder, where the affective symptoms are serious enough to negatively impact functioning
(Takeda et al., 2022). Thus, having such a high number of people with PMDD can create the
question of whether this study included a clinical sample. Further, research shows that people
with PMDD have heightened subjective stress, especially in their late luteal phase (Beddig et al.,
2019). There also is evidence that people with PMDD display altered basal HPA activity (Beddig
et al., 2019). Indeed, the current study showed that the group with moderate to severe PMS or
PMDD also showed higher levels of chronic stress compared to the rest of the sample. This
group also differed from the rest of the sample in menstrual beliefs, having stronger beliefs in
menstruation being disabling and annoying. While it may not directly influence the results of this
study, this proportion of participants with moderate to severe PMS or PMDD may make these
results less generalizable to other populations.
The means from this sample for the different variables considered are comparable to
other studies, with the exception of the PSS scores and salivary progesterone levels. The mean
PSS score from the current study was 29.93, whereas previous studies show means of 23.18 to
23.67 (Cohen et al., 1983). This higher mean of PSS scores may be due to the number of
participants with moderate to severe PMS or PMDD in this sample. This study sample also had
higher levels of progesterone compared to previous studies. For example, a previous study
testing participants at different points of the luteal phase showed mean salivary progesterone

84
concentrations of 73 pg/mL (Delfs et al., 1994), whereas the current study had a mean of 300.85
pg/mL in the luteal phase. It is possible that assay-specific issues may have created this
difference. When running the immunoassay, positive controls were provided by the
manufacturer, containing a specific concentration of estradiol and progesterone. These controls
were run to ensure that the assay correctly detected the hormones within the passive drool
samples. Interestingly, the results of the positive control also showed higher progesterone
concentrations than expected, suggesting the assay itself may have been responsible for the
higher concentrations. Further, duplicate assay plates were run to ensure that hormone levels did
not differ significantly between the plates. The values resulting from these duplicates were
similar, supporting that these higher progesterone levels were due to an assay-specific issue. One
reason why this assay-specific issue arose may be the method of washing the plates. When
conducting immunoassays, the plates in which the hormones and reagents are added must be
washed in between samples to remove the competitive reagent used in the immunoassay. In this
study, it is possible that these plates were not washed thoroughly enough and may have leftover
traces of the competitive reagent. Thus, the difference in the progesterone levels seen in this
study compared to other studies is likely assay specific instead of an actual difference in this
sample. While the higher progesterone levels may be a methodological issue, the heightened
chronic stress in this sample may make these findings less generalizable.
The results did not show a difference in salivary progesterone and estradiol across the
menstrual cycle. The distributions of hormone levels per cycle phase were graphed and visually
explored to determine if hormonal levels across the phases seemed to be less typical for
participants with low-confidence phase estimations. While this low-confidence phase estimation
could not be systematically assigned for all participants with non “typical” hormone levels, it

85
was true for some of the participants. Thus, the manner by which the cycle phase was assigned
was not stringent enough, creating variation in the hormone levels. The presence of participants
with non “typical” hormone levels as well as low confidence phase estimations demonstrate the
importance of computing these confidence scores to better understand the data. Further, the low
confidence scores also indicate that the method of tracking participants used in the Winter 2023
semester was superior; although it was demanding, the participants were more likely to be tested
in the cycle phases of interest. Another reason this study found a lack of difference in hormone
levels across the menstrual cycle could also be due to a lack of power in this study. As such, with
data collection continuing, it is possible that the typical ovarian hormone fluctuations across the
menstrual cycle may arise. Thus, the lack of typical ovarian hormone fluctuations could be a
result of methodological or sample size-related issues.
5.2 Effect of Menstrual Cycle Phase and Ovarian Hormones on Learning Processes
The results showed that there was no effect of the menstrual cycle phase on the learning
processes that were used during the probabilistic classification task. While this lack of effect may
be due to the low power, it could also indicate that the learning processes are affected more by
the dimensions of the menstrual cycle instead of the phases. Indeed, the current study found a
trend for the main effect of progesterone on learning processes. Specifically, greater
concentrations of salivary progesterone were related to a greater cognitive bias. This finding
corroborates what was reported by Hussain and colleagues (2016). Unlike the current study,
Hussain and colleagues found a difference in the learning processes used in a spatial learning
task across the menstrual cycle phases. Similar to the current study, they also found trending
differences in hormone levels between people who were categorized as using cognitive and
habitual learning processes; specifically, there were higher progesterone and lower estradiol

86
levels in people using cognitive learning processes. However, Hussain and colleagues (2016)
categorized the participants as being either cognitive or habitual learners, whereas the current
study only had enough power to look at the changes in strategy dominance scores (i.e., the extent
to which one strategy was used more than the other). Since Hussain and colleagues (2016) did
not compute strategy dominance scores, it was not possible to make full comparisons between
the results of both studies. Contrastingly, Scheuringer and Pletzer (2017) found the opposite
results, where higher levels of progesterone were related to higher accuracy when using the
egocentric (or habitual) learning processes. However, the study by Hussain and colleagues
(2016) is more relevant to the current study in that both studies allowed participants to use their
preferred strategy, whereas Scheuringer and Pletzer (2017) instructed them to use a specific
strategy. Thus, the current findings of no effect of menstrual cycle phase and a trend association
with progesterone are partially replicating previous relevant literature.
This effect of progesterone on learning processes can be facilitated through the brain
regions previously implicated in the recruitment of learning processes. For example, Hussain and
colleagues (2016) suggested that the combined effects of progesterone and estradiol on the
hippocampus could underlie their findings. Specifically, the authors proposed that estradiol may
act on hippocampal activity and thus cognitive learning in an inverted-U manner, where optimal
effects arise with moderate levels of estradiol and any further reductions or increases in estradiol
can inhibit cognitive learning. However, the detrimental effects of the luteal peak of estradiol on
cognitive learning can be reversed with the accompanying release of progesterone (Hussain et
al., 2016). Interestingly, a rise in habitual learning in the follicular phase with higher levels of
estradiol would be expected with this explanation, which was not seen in the current study. This
lack of effect is due to the interaction of estradiol and chronic stress seen in this study. Thus, the

87
preliminary findings of a trend association between progesterone and strategy dominance scores
may be understood by progesterone’s action in the hippocampus.
5.3 Effect of Stress on Learning Processes
The results of this study showed that within a sample of naturally-cycling women, there
is no correlation between the levels of subjective chronic stress and the use of learning processes
in a probabilistic classification learning task. Generally, most previous studies show a
relationship between stress and learning processes, such that the presence of stress creates a bias
toward habitual learning (Wirz et al., 2018). However, these studies mostly focus on acute stress.
There are some studies investigating the effects of chronic stress on learning processes that
measure stress using the Perceived Stress Scale, and these studies show mixed findings. For
example, Bohbot and colleagues (2011) investigated the use of learning processes in a spatial
learning task similar to the one used by Hussain and colleagues (2016). They also measured
salivary cortisol and reported chronic stress. The results were contradictory to what other studies
consistently found, in that there were greater cortisol concentrations in participants who
preferred using the cognitive learning process. Further, the study also found no differences
between people categorized as using cognitive or habitual processes in terms of PSS scores,
similar to the current study. Another study by Lenow and colleagues (2017) looked at the
association between acute and chronic stress and the use of cognitive and habitual learning
processes within decision-making. Specifically, the participants completed a computer task after
being exposed to an acute stressor. In the task, they were presented with a tree on the screen and
were asked to either pick apples from the current tree (exploitative or habitual process) or move
to the next tree (explorative or cognitive process). The goal for the participants was to pick as
many apples as possible. The participants completed the PSS prior to exposure to the acute

88
stressor and provided salivary cortisol samples throughout the study. Unlike the study by Bohbot
and colleagues (2011), the results showed that participants engaged in more exploitation with
higher levels of cortisol and higher chronic stress as reported by the PSS. As such, there are
mixed findings in terms of how PSS scores correlate with learning processes. Of note, the
samples for both of these studies included female participants, without restricting their menstrual
cycle phase. So, it is possible that there were interactive effects of stress and ovarian hormones
that could explain the inconsistency in the results.
While the present study did not find an effect of the PSS score on strategy dominance
scores, we did find associations with variations in menstrual-related attitudes and beliefs. Thus, it
is possible that the stress participants experienced may not have been encapsulated by the items
presented on the PSS. While the items on the PSS can capture standard stressful situations, it
misses out on situations that are implicitly stressful, such as dealing with the stigma around
menstruation. For example, there is research showing that having greater awareness of the stigma
around one’s identity can influence how one physiologically reacts to social stressors (Doyle &
Molix, 2018). This study highlights that while the PSS could capture the standard stressful
situations people face in their lives, it cannot capture the nuance of menstrual stigma and
stereotypes and how these factors may influence the stress menstruating individuals face.
Further, half of the current sample was categorized as having either moderate to severe PMS or
PMDD, and these participants also created the entire group labelled as high stress in this study.
The experiences that these participants have regarding their menstrual cycle can be a dominant
source of stress, which is another factor that is not accounted for by the PSS. As such, it is
possible that while there was no correlation between the PSS scores and learning processes

89
within this sample of naturally cycling women, there may be stressors related to menstruation or
the menstrual cycle that may modulate the use of learning processes.
Constructs that were related to the use of learning processes included beliefs of
menstruation being bothersome or something to be kept secret. Specifically, the stronger these
beliefs were in participants, the more habitual bias they showed during the WPT. Keeping
menstruation a secret can be a stressful experience in itself. Menstruating people tend to feel
more self-conscious and try to hide signs of menstruation to ensure that people around them are
not able to see their menstrual status (Johnston-Robledo & Chrisler, 2013). Studies show that
people tend to view those who show signs of menstruation (i.e., through seeing an unopened
tampon in their possession) as unlikeable and incompetent (Roberts et al., 2002). Thus, hiding
menstruation can be very important to menstruating people and can potentially create stress in
their lives due to the possible consequences of having their menstrual status revealed. This
underlying stress may be a factor that contributes to this finding of stronger beliefs in secrecy
being related to a habitual bias. Further, experiencing menstruation as a bothersome event can
also be a factor that modulates how people experience stress. A study by Meng and colleagues
(2022) revealed that beliefs in menstruation being a bothersome event can mediate the
relationship between PMS and people’s perceptions of their physiological stress responses
during their menstruation. As such, the relationship between the annoyance towards
menstruation and people’s subjective experience of their physiological stress response reinforces
the association between the social construction of the menstrual cycle and stress. Similar to the
study by Doyle and Molix (2018) showing a blunted activity of the HPA axis in people who
were more aware of the stigma related to their gender, the participants in this current study may
also show changes in their cortisol profile related to the beliefs they hold regarding menstruation.

90
Thus, how people conceptualize and articulate the menstrual cycle is tied to a broader set of
experiences, which can then have percolating effects on cognitive and habitual learning
processes.
As such, how women experience the menstrual cycle can be a better predictor of learning
processes than chronic stress itself. Specifically, items in the bothersome and secrecy subscales
could be a proxy of stress experiences. For example, the bothersome subscale of the MAQ
includes items like “Menstruation is just something I have to put up with,” placing the menstrual
cycle in a negative light and describing it as an annoyance (Brooks-Gunn & Ruble, 1980).
Further, the secrecy subscale of the BATM includes items such as “It is embarrassing when a
man finds out that a woman is having her period,” highlighting how it can be stressful for the
menstruating person if someone finds out about their menses (Marván et al., 2006). As such, it is
possible that menstrual beliefs and attitudes reflect untapped stress dimensions, which could
underlie the correlations found in the present study.
5.4 Interactive Effects of Stress and Ovarian Hormones on Learning Processes
A central prediction of this study was that the variation in learning processes would be
better predicted by looking at the interaction between stress and ovarian hormones, and the
preliminary findings support this prediction. Specifically, the results show an interaction effect of
salivary estradiol and PSS scores on strategy dominance scores. For participants with low levels
of reported chronic stress, higher salivary estradiol concentrations were associated with greater
cognitive bias. The opposite was true for participants with high levels of reported chronic stress,
where higher salivary estradiol concentrations were associated with greater habitual bias. These
findings indicate that estradiol functioned to emphasize the participant’s context and facilitate the
adaptation to stress. The finding partially contradicts what was found in the study by McHale

91
(2019) that investigated the interactive effect between ovarian hormones and stress on spatial
learning processes. In this study, there were no individual or interactive effects of acute stress
and the menstrual cycle phase on the use of learning processes in spatial navigation. The study
however did not include hormonal measures, which is where the current study was able to find
effects. Further, the current study also looked at measures of chronic stress instead of acute
stress. This interactive effect may be operating on a longer timescale that is captured by
measuring chronic stress instead of acute stress. Thus, the study by McHale (2019) may have
been unable to find this interactive effect due to the measures that were used. Moreover,
considering the complexity of the menstrual cycle arising from its biopsychosocial
characterization, it is possible that measuring salivary ovarian hormone levels may be more
informative in such studies than using the menstrual cycle phase as a proxy. In sum, the current
study did somewhat replicate the findings of the previous study that looked at similar interactive
effects and was able to expand on the findings by adding nuances from the hormonal findings
and reported chronic stress.
This interaction may in part be due to the dopaminergic system, which influences
learning processes and is affected by stress and estradiol. The dopaminergic system, which is
responsible for reward learning, underlies many facets of cognitive and habitual learning
(Goodman, 2021). Specifically, one study in humans shows that greater dopamine levels in the
ventral striatum (a region underlying both cognitive and habitual learning; Huang et al., 2020;
McDannald et al., 2012) can lead to a cognitive bias and can change how cognitive learning is
signalled in the lateral prefrontal cortex and habitual learning is signalled in the ventral striatum
(Deserno et al., 2015). Importantly, the dopaminergic system can be influenced by the
experience of chronic stress. For instance, a study by Bloomfield and colleagues (2019) showed

92
that chronic psychosocial stress can reduce dopamine synthesis within the ventral striatum.
Further, this dopaminergic system can interact with estradiol levels, in that the effects of
estradiol on cognitive functions are clearer when accounting for dopamine levels (Jacobs &
D’Esposito, 2011). Specifically, Jacobs and D’Esposito (2011) found that during a working
memory task, people with lower dopamine levels in the prefrontal cortex had greater
performance during a high estradiol state and people with high dopamine had greater
performance during a low estradiol state. (Jacobs & D’Esposito, 2011). As the dopaminergic,
stress, and reproductive systems interact with one another within brain regions related to the
learning processes and are seen to influence learning processes individually, it is plausible that
the interactive effect of estradiol and chronic stress can be mediated by the dopaminergic system.
5.5 Limitations and Future Directions
The present study had multiple limitations that can be addressed or considered when
conducting similar studies or interpreting the present results. Firstly, the preliminary findings of
this study resulted from a small sample size. As such, the current results may disappear or further
effects may be found with continued testing. Moreover, the small sample size limited the types
of analyses possible as it did not permit analyses using ANOVA or more complex regression
models. Second, the initial phase of recruitment resulted in greater attrition. During this phase,
many participants either stopped responding to the physiological surveys, did not book study
sessions, or did not attend their booked sessions. This level of attrition indicates that the initial
method of recruitment imposed a greater workload on the participants. This workload was
significantly decreased when participants were asked to book their sessions right away.
However, booking the sessions right away meant the testing sessions took place across the

93
menstrual cycle instead of the phases of interest. Thus, there exists a compromise here of
whether sample size or cycle phase assignment holds more importance.
Some limitations arose from the menstrual cycle phase estimation. Specifically, it was
difficult to accurately determine which phase participants were tested in if they missed multiple
physiological questionnaires in a row. For example, there was no menstrual cycle information for
four days if the participants missed two physiological questionnaires in a row, which could
possibly encompass an entire menstrual period. Thus, studies using a similar tracking method
may benefit from incentivizing the completion of as many physiological questionnaires as
possible. Another issue with the cycle phase estimation was participants guessing the menstrual
cycle context of the study based on the physiological survey questions. Some participants noted
that the questions reflected their symptoms during their menses. Thus, it is important to find
other questions regarding physiology that don’t clue the participants into the menstrual cycle
context of the study. Despite these two limitations, this study still used a rigorous manner of
tracking participants while keeping them blinded to the study’s purpose.
For future directions, participant testing will continue in order to reach the target sample
size. Reaching the target sample size will increase the power and allow more complex analyses,
such as regression models with three or more variables or ANOVA. In addition, this study will
incorporate another analysis done by Zerbes and colleagues (2020), investigating how influences
of contextual factors on strategy dominance scores, preference of learning processes, and
learning performance change across 25-trial intervals, instead of looking at the entire task
altogether. This analysis will determine if dimensions of the menstrual cycle or chronic stress
can influence this progression across the task. As such, recruiting more participants will both
increase the study’s power and allow more informative data analyses.

94
Future studies can better understand how stress and menstrual experiences influence
learning processes by incorporating more diverse manners of capturing menstrual experiences.
For example, half of the current sample experienced moderate to severe PMS or PMDD, and
these participants also had higher levels of chronic stress compared to the rest of the sample.
Thus, these participants experienced a unique source of stress that cannot be encapsulated fully
by the instruments included in the current study. As such, the Menstrual Distress Questionnaire
(MDQ) by Moos (1968) would be a valuable addition to studies investigating menstrual
experiences and beliefs as dimensions of stress. This questionnaire asks participants to indicate
the extent to which they experience various menstrual-related symptoms, and evidence points to
MDQ scores being a correlate of subjective stress levels (Matsumoto et al., 2019). Thus, adding
the MDQ would quantify menstrual experiences and help analyze how these experiences can
play a role in learning processes. In addition to the MDQ, future studies would benefit from
adding diverse measures of menstrual beliefs. Indeed, there is evidence of these beliefs differing
based on various demographic measures (i.e., religion, nationality, gender identity; Chrisler et
al., 2016; Kaundal & Thakur, 2014; Spadaro et al., 2018). As such, the measures currently
available to capture menstrual beliefs may be too constrained, and the incorporation of
interviews to further understand people’s menstrual beliefs may be more informative. Thus,
future studies can incorporate interviews and the MDQ to better understand how people view and
experience their menstrual cycle, providing a more complex picture of the role that the menstrual
cycle plays in learning processes.
6. Conclusion
One critical component of understanding how we navigate the environment, predict
outcomes, or make decisions is to determine the role of context. This learning can occur through

95
cognitive or habitual processes, and various contextual factors can influence which learning
processes are recruited (Packard & Goodman, 2013). While preliminary, the results showed that
while the menstrual cycle phase does not influence the use of learning processes, there was a
trend association between salivary progesterone and the use of learning processes. Further, the
study revealed that there is an interaction between salivary estradiol and chronic stress in
predicting the use of learning processes. Lastly, this study provides evidence that the use of
learning processes is also related to the beliefs people hold regarding menstruation. As such, this
study showed that there is a complex biopsychosocial context surrounding the menstrual cycle,
and the different dimensions of this context can interact to influence the manner in which
naturally-cycling people learn. The results then emphasize the value in examining the interactive
effect of stress and ovarian hormones instead of viewing them as individual modulating factors.
In doing so, this study also highlights the importance of viewing the menstrual cycle as
multidimensional instead of focusing on the hormonal dimension alone.

96
References
Akam, T., Rodrigues-Vaz, I., Marcelo, I., Zhang, X., Pereira, M., Oliveira, R. F., Dayan, P., &
Costa, R. M. (2021). The Anterior Cingulate Cortex Predicts Future States to Mediate
Model-Based Action Selection. Neuron, 109(1), 149–163.
https://doi.org/10.1016/j.neuron.2020.10.013
Barel, E., Abu-Shkara, R., Colodner, R., Masalha, R., Mahagna, L., Zemel, O. C., & Cohen, A.
(2018). Gonadal hormones modulate the HPA-axis and the SNS in response to
psychosocial stress. Journal of Neuroscience Research, 96(8), 1388–1397.
https://doi.org/10.1002/jnr.24259
Bayer, J., Rusch, T., Zhang, L., Gläscher, J., & Sommer, T. (2020). Dose-dependent effects of
estradiol on prediction error related neural activity in the nucleus accumbens of healthy
young women. Psychopharmacology, 237(3), 745–755. https://doi.org/10.1007/s00213019-05409-7
Beddig, T., Reinhard, I., & Kuehner, C. (2019). Stress, mood, and cortisol during daily life in
women with Premenstrual Dysphoric Disorder (PMDD). Psychoneuroendocrinology,
109, 104372.
Beltz, A. M., & Moser, J. S. (2020). Ovarian hormones: A long overlooked but critical
contributor to cognitive brain structures and function. Annals of the New York Academy
of Sciences, 1464(1), 156–180. https://doi.org/10.1111/nyas.14255
Berner, L. A., & Marsh, R. (2014). Frontostriatal Circuits and the Development of Bulimia
Nervosa. Frontiers in Behavioral Neuroscience, 8, 395.
https://doi.org/10.3389/fnbeh.2014.00395

97
Bloomfield, M. A., McCutcheon, R. A., Kempton, M., Freeman, T. P., & Howes, O. (2019). The
effects of psychosocial stress on dopaminergic function and the acute stress response.
ELife, 8, e46797. https://doi.org/10.7554/eLife.46797
Bohbot, V. D., Gupta, M., Banner, H., & Dahmani, L. (2011). Caudate nucleus-dependent
response strategies in a virtual navigation task are associated with lower basal cortisol
and impaired episodic memory. Neurobiology of Learning and Memory, 96(2), 173-180.
Brake, W. G., & Lacasse, J. M. (2018). Sex differences in spatial navigation: The role of gonadal
hormones. Current Opinion in Behavioral Sciences, 23, 176–182.
https://doi.org/10.1016/j.cobeha.2018.08.002
Brooks-Gunn, J., & Ruble, D. N. (1980). The menstrual attitude questionnaire. Psychosomatic
Medicine, 42(5), 503–512.
Castro, R. T., Ehlert, U., & Fischer, S. (2021). Variation in genes and hormones of the
hypothalamic-pituitary-ovarian axis in female mood disorders - a systematic review and
meta-analysis. Frontiers in Neuroendocrinology, 62, 100929–100929.
https://doi.org/10.1016/j.yfrne.2021.100929
Chrisler, J. C., Gorman, J. A., Manion, J., Murgo, M., Barney, A., Adams-Clark, A., Newton, J.
R., & McGrath, M. (2016). Queer periods: Attitudes toward and experiences with
menstruation in the masculine of centre and transgender community. Culture, Health &
Sexuality, 18(11), 1238–1250. https://doi.org/10.1080/13691058.2016.1182645
Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress.
Journal of Health and Social Behaviour, 24, 385-396. https://doi.org/10.2307/2136404
Coughlin, P. C. (1990). Premenstrual syndrome: How marital satisfaction and role choice affect
symptom severity. Social Work, 35(4), 351-355.

98
Daw, N. D., Niv, Y., & Dayan, P. (2005). Uncertainty-based competition between prefrontal and
dorsolateral striatal systems for behavioral control. Nature Neuroscience, 8(12), 1704–
1711. https://doi.org/10.1038/nn1560
De Leo, V., Musacchio, M. C., Cappelli, V., Massaro, M. G., Morgante, G., & Petraglia, F. J. R.
B. (2016). Genetic, hormonal and metabolic aspects of PCOS: An update. Reproductive
Biology and Endocrinology, 14(1), 1-17.
Delfs, T. M., Klein, S., Fottrell, P., Naether, O. G., Leidenberger, F. A., & Zimmermann, R. C.
(1994). 24-hour profiles of salivary progesterone. Fertility and Sterility, 62(5), 960-966.
Deserno, L., Huys, Q. J. M., Boehme, R., Buchert, R., Heinze, H.-J., Grace, A. A., Dolan, R. J.,
Heinz, A., & Schlagenhauf, F. (2015). Ventral striatal dopamine reflects behavioral and
neural signatures of model-based control during sequential decision making. Proceedings
of the National Academy of Sciences, 112(5), 1595–1600.
https://doi.org/10.1073/pnas.1417219112
Diekhof, E. K., Geana, A., Ohm, F., Doll, B. B., & Frank, M. J. (2021). The straw that broke the
camel’s back: Natural variations in 17β-estradiol and COMT-Val158Met genotype
interact in the modulation of model-free and model-based control. Frontiers in
Behavioral Neuroscience, 15, 658769. https://doi.org/10.3389/fnbeh.2021.658769
Dixon, M. L., & Christoff, K. (2014). The lateral prefrontal cortex and complex value-based
learning and decision making. Neuroscience and Biobehavioral Reviews, 45, 9–18.
https://doi.org/10.1016/j.neubiorev.2014.04.011
Doody, M., Van Swieten, M. M. H., & Manohar, S. G. (2022). Model-based learning
retrospectively updates model-free values. Scientific Reports, 12(1), 2358.
https://doi.org/10.1038/s41598-022-05567-3

99
Doyle, D. M., & Molix, L. (2018). Stigma consciousness modulates cortisol reactivity to social
stress in women. European Journal of Social Psychology, 48(2), 217-224.
Epstein, R. A., Patai, E. Z., Julian, J. B., & Spiers, H. J. (2017). The cognitive map in humans:
Spatial navigation and beyond. Nature Neuroscience, 20, 1504-1513.
https://doi.org/10.1038/nn.4656
Gabrys, R. L., Tabri, N., Anisman, H., & Matheson, K. (2018). Cognitive control and flexibility
in the context of stress and depressive symptoms: The cognitive control and flexibility
questionnaire. Frontiers in Psychology, 9, 2219.
https://doi.org/10.3389/fpsyg.2018.02219
Gahnstrom, C. J., & Spiers, H. J. (2020). Striatal and hippocampal contributions to flexible
navigation in rats and humans. Brain and Neuroscience Advances, 4, 1–7.
https://doi.org/10.1177/2398212820979772
Gandara, B. K., Leresche, L., & Mancl, L. (2007). Patterns of salivary estradiol and progesterone
across the menstrual cycle. Annals of the New York Academy of Sciences, 1098(1), 446450.
Gillan, C. M., & Robbins, T. W. (2014). Goal-directed learning and obsessive–compulsive
disorder. Philosophical Transactions of the Royal Society B: Biological Sciences,
369(1655), 20130475. https://doi.org/10.1098/rstb.2013.0475
Gläscher, J., Daw, N., Dayan, P., & O’Doherty, J. P. (2010). States versus Rewards: Dissociable
Neural Prediction Error Signals Underlying Model-Based and Model-Free Reinforcement
Learning. Neuron, 66(4), 585–595. https://doi.org/10.1016/j.neuron.2010.04.016

100
Gluck, M. A., Shohamy, D., & Myers, C. (2002). How do people solve the “weather prediction”
task?: Individual variability in strategies for probabilistic category learning. Learning &
Memory, 9(6), 408–418. https://doi.org/10.1101/lm.45202
Gomez-Perales, E. L., & Brake, W. G. (2022). The role of progesterone in memory bias during
spatial navigation in females. Journal of Neuroendocrinology, 35, e13197.
https://doi.org/10.1111/jne.13197
Goodman, J., & Packard, M. G. (2016). Memory Systems and the Addicted Brain. Frontiers in
Psychiatry, 7, 24. https://doi.org/10.3389/fpsyt.2016.00024
Goodman, J. (2021). Place vs. response learning: History, controversy, and neurobiology.
Frontiers in Behavioral Neuroscience, 14, 598570.
https://doi.org/10.3389/fnbeh.2020.598570
Haggerty, C. L., Ness, R. B., Kelsey, S., & Waterer, G. W. (2003). The impact of estradiol and
progesterone on asthma. Annals of Allergy, Asthma, & Immunology, 90(3), 284-291.
https://doi.org/10.1016/S1081-1206(10)61794-2
Hampson, E. (2020). A brief guide to the menstrual cycle and oral contraceptive use for
researchers in behavioral endocrinology. Hormones and Behavior, 119, 104655.
Hara, Y., Waters, E. M., McEwen, B. S., & Morrison, J. H. (2015). Estrogen Effects on
Cognitive and Synaptic Health Over the Lifecourse. Physiological Reviews, 95(3), 785–
807. https://doi.org/10.1152/physrev.00036.2014
Harms, M. B. (2017). Stress and exploitative decision-making. The Journal of Neuroscience, 37,
10035-10037. https://doi.org/10.1523/JNEUROSCI.2169-17.2017
Harris, T., Scheuringer, A., & Pletzer, B. (2019). Perspective and strategy interactively modulate
sex differences in a 3D navigation task. Biology of Sex Differences, 10(17), 1-12.

101
Hilz, E. N. (2022). Methods and considerations for the use of hormonal contraceptives in rat
models of neurobehavior. Frontiers in Neuroendocrinology, 66, 101011.
https://doi.org/10.1016/j.yfrne.2022.101011
Huang, Y., Yaple, Z. A., & Yu, R. (2020). Goal-oriented and habitual decisions: Neural
signatures of model-based and model-free learning. NeuroImage, 215, 116834.
https://doi.org/10.1016/j.neuroimage.2020.116834
Hussain, D., Hanafi, S., Konishi, K., Brake, W.G., & Bohbot, V.D. (2016). Modulation of spatial
and response strategies by phase of the menstrual cycle in women tested in a virtual
navigation task. Psychoneuroendocrinology, 70, 108-117.
https://doi.org/10.1016/j.psyneuen.2016.05.008
Jacobs, E., & D’Esposito, M. (2011). Estrogen shapes dopamine-dependent cognitive processes:
Implications for women’s health. The Journal of Neuroscience, 31(14), 5286–5293.
https://doi.org/10.1523/JNEUROSCI.6394-10.2011
Joëls, M., Pu, Z., Wiegert, O., Oitzl, M. S., & Krugers, H. J. (2006). Learning under stress: How
does it work? Trends in Cognitive Sciences, 10(4), 152–158.
https://doi.org/10.1016/j.tics.2006.02.002
Johnston-Robledo, I., & Chrisler, J. C. (2020). The menstrual mark: Menstruation as social
stigma. The Palgrave Handbook of Critical Menstruation Studies, 181-199.
Jukic, A. M. Z., Weinberg, C. R., Wilcox, A. J., McConnaughey, D. R., Hornsby, P., & Baird, D.
D. (2008). Accuracy of reporting of menstrual cycle length. American Journal of
Epidemiology, 167(1), 25-33.

102
Kalra, P. B., Gabrieli, J. D. E., & Finn, A. S. (2019). Evidence of stable individual differences in
implicit learning. Cognition, 190, 199–211.
https://doi.org/10.1016/j.cognition.2019.05.007
Kaundal, M., & Thakur, B. (2014). A dialogue on menstrual taboo. Indian Journal of Community
Health, 26(2), 192-195.
Knudsen, U. B., Tabor, A., Mosgaard, B., Andersen, E. S., Kjer, J. J., Hahn-Pedersen, S., ... &
Mogensen, O. (2004). Management of ovarian cysts. Acta obstetricia et gynecologica
Scandinavica, 83(11), 1012-1021.
Korol, D. L., & Kolo, L. L. (2002). Estrogen-induced changes in place and response learning in
young adult female rats. Behavioral Neuroscience, 116(3), 411–420.
https://doi.org/10.1037/0735-7044.116.3.411
Korol, D.L., Malin, E.L., Borden, K.A., Busby, R.A., & Couper-Leo, J. (2004). Shifts in
preferred learning strategy across the estrous cycle in female rats. Hormones and
Behaviour, 45, 330-338. https://doi.org/10.1016/j.yhbeh.2004.01.005
Kung, F. Y., Kwok, N., & Brown, D. J. (2018). Are attention check questions a threat to scale
validity?. Applied Psychology, 67(2), 264-283.
Lacasse, J. M., Patel, S., Bailey, A., Peronace, V., & Brake, W. G. (2022). Progesterone rapidly
alters the use of place and response memory during spatial navigation in female rats.
Hormones and Behavior, 140, 105137. https://doi.org/10.1016/j.yhbeh.2022.105137
Le, J., Thomas, N., & Gurvich, C. (2020). Cognition, the menstrual cycle, and premenstrual
disorders: A review. Brain Sciences, 10(4), 198.
Lemhöfer, K., & Broersma, M. (2012). Introducing LexTALE: A quick and valid lexical test for
advanced learners of English. Behavior Research Methods, 44, 325-343.

103
Lenow, J. K., Constantino, S. M., Daw, N. D., & Phelps, E. A. (2017). Chronic and acute stress
promote overexploitation in serial decision making. The Journal of Neuroscience, 37(23),
5681–5689. https://doi.org/10.1523/JNEUROSCI.3618-16.2017
Lewis, C. A., Kimmig, A.-C. S., Kroemer, N. B., Pooseh, S., Smolka, M. N., Sacher, J., &
Derntl, B. (2022). No Differences in Value-Based Decision-Making Due to Use of Oral
Contraceptives. Frontiers in Endocrinology, 13, 817825.
https://doi.org/10.3389/fendo.2022.817825
Marcondes, F. K., Bianchi, F. J., & Tanno, A. P. (2002). Determination of the estrous cycle
phases of rats: Some helpful considerations. Brazilian Journal of Biology, 62(4A), 609614. https://doi.org/10.1590/s1519-69842002000400008
Martel, J. C., & Gatti McArthur, S. (2020). Dopamine Receptor Subtypes, Physiology and
Pharmacology: New Ligands and Concepts in Schizophrenia. Frontiers in
Pharmacology, 11, 1003. https://doi.org/10.3389/fphar.2020.01003
Marván, M. L., Cortés-Iniestra, S., & González, R. (2005). Beliefs about and attitudes toward
menstruation among young and middle-aged Mexicans. Sex Roles, 53(3–4), 273–279.
https://doi.org/10.1007/s11199-005-5685-3
Marván, Ma. L., Ramírez-Esparza, D., Cortés-Iniestra, S., & Chrisler, J. C. (2006). Development
of a new scale to measure beliefs about and attitudes toward menstruation (BATM): Data
from Mexico and the United States. Health Care for Women International, 27(5), 453–
473. https://doi.org/10.1080/07399330600629658
Marván, Ma. L., Vázquez-Toboada, R., & Chrisler, J. C. (2014). Ambivalent sexism, attitudes
towards menstruation and menstrual cycle-related symptoms. International Journal of
Psychology, 49(4), 280–287. https://doi.org/10.1002/ijop.12028

104
Matsumoto, T., Egawa, M., Kimura, T., & Hayashi, T. (2019). A potential relation between
premenstrual symptoms and subjective perception of health and stress among college
students: A cross-sectional study. BioPsychoSocial Medicine, 13(1), 26.
https://doi.org/10.1186/s13030-019-0167-y
McDannald, M. A., Takahashi, Y. K., Lopatina, N., Pietras, B. W., Jones, J. L., & Schoenbaum,
G. (2012). Model-based learning and the contribution of the orbitofrontal cortex to the
model-free world. European Journal of Neuroscience, 35(7), 991–996.
https://doi.org/10.1111/j.1460-9568.2011.07982.x
McHale, A. L. (2019). Effects of menstrual cycle phase and stress on spatial navigational
strategy use [Dissertation]. University of Cape Town.
Meng, Y., Chang, L., Hou, L., & Zhou, R. (2022). Menstrual attitude and social cognitive stress
influence autonomic nervous system in women with premenstrual syndrome. Stress,
25(1), 87–96. https://doi.org/10.1080/10253890.2021.2024163
Moos, R. H. (1968). The development of a Menstrual Distress Questionnaire. Psychosomatic
Medicine, 30(6), 853–867. https://doi.org/10.1097/00006842-196811000-00006
Noachtar, I., Harris, T.-A., Hidalgo-Lopez, E., & Pletzer, B. (2022). Sex and strategy effects on
brain activation during a 3D-navigation task. Communications Biology, 5, 234.
https://doi.org/10.1038/s42003-022-03147-9
O’Doherty, J. P., Cockburn, J., & Pauli, W. M. (2017). Learning, reward, and decision making.
Annual Review of Psychology, 68, 73–100. https://doi.org/10.1146/annurev-psych010416-044216
Packard, M. G., & Goodman, J. (2013). Factors that influence the relative use of multiple
memory systems. Hippocampus, 23(11), 1044–1052. https://doi.org/10.1002/hipo.22178

105
Patterson, T. K., & Knowlton, B. J. (2018). Subregional specificity in human striatal habit
learning: A meta-analytic review of the fMRI literature. Current Opinion in Behavioral
Sciences, 20, 75–82. https://doi.org/10.1016/j.cobeha.2017.10.005
Phumsatitpong, C., Wagenmaker, E. R., & Moenter, S. M. (2021). Neuroendocrine interactions
of the stress and reproductive axes. Frontiers in Neuroendocrinology, 63, 100928.
https://doi.org/10.1016/j.yfrne.2021.100928
Quinlan, M. G., Hussain, D., & Brake, W. G. (2008). Use of cognitive strategies in rats: The role
of estradiol and its interaction with dopamine. Hormones and Behavior, 53(1), 185–191.
https://doi.org/10.1016/j.yhbeh.2007.09.015
Radenbach, C., Reiter, A. M. F., Engert, V., Sjoerds, Z., Villringer, A., Heinze, H.-J., Deserno,
L., & Schlagenhauf, F. (2015). The interaction of acute and chronic stress impairs modelbased behavioral control. Psychoneuroendocrinology, 53, 268–280.
https://doi.org/10.1016/j.psyneuen.2014.12.017
Roberts, T.-A., Goldenberg, J. L., Power, C., & Pyszczynski, T. (2002). “Feminine protection”:
The effects of menstruation on attitudes towards women. Psychology of Women
Quarterly, 26, 131–139. https://doi.org/10.1111/1471-6402.00051
Romans, S., Clarkson, R., Einstein, G., Petrovic, M., & Stewart, D. (2012). Mood and the
menstrual cycle: A review of prospective data studies. Gender Medicine, 9(5), 361-384.
https://doi.org/10.1016/j.genm.2012.07.003
Sarrel, P. M., Sullivan, S. D., & Nelson, L. M. (2016). Hormone replacement therapy in young
women with surgical primary ovarian insufficiency. Fertility and Sterility, 106(7), 15801587.

106
Scheuringer, A., & Pletzer, B. (2017). Sex differences and menstrual cycle dependent changes in
cognitive strategies during spatial navigation and verbal fluency. Frontiers in
Psychology, 8, 1-12. https://doi.org/10.3389/fpsyg.2017.00381
Schmalenberger, K. M., Tauseef, H. A., Barone, J. C., Owens, S. A., Lieberman, L., Jarczok, M.
N., Girdler, S. S., Kiesner, J., Ditzen, B., & Eisenlohr-Moul, T. A. (2021). How to study
the menstrual cycle: Practical tools and recommendations. Psychoneuroendocrinology,
123, 104895. https://doi.org/10.1016/j.psyneuen.2020.104895
Schmidt, P. J., Martinez, P. E., Nieman, L. K., Koziol, D. E., Thompson, K. D., Schenkel, L.,
Wakim, P. G., & Rubinow, D. R. (2017). Exposure to a change in ovarian steroid levels
but not continuous stable levels triggers PMDD symptoms following ovarian
suppression. American Journal of Psychiatry, 174(10), 980–989.
https://doi.org/10.1176/appi.ajp.2017.16101113
Schoenberg, H. L., Bremer, G. P., Carasi-Schwartz, F., VonDoepp, S., Arntsen, C., Anacker, A.
M. J., & Toufexis, D. J. (2022). Cyclic estradiol and progesterone during instrumental
acquisition contributes to habit formation in female rats. Hormones and Behavior, 142,
105172. https://doi.org/10.1016/j.yhbeh.2022.105172
Schwabe, L., Oitzl, M. S., Philippsen, C., Richter, S., Bohringer, A., Wippich, W., &
Schachinger, H. (2007). Stress modulates the use of spatial versus stimulus-response
learning strategies in humans. Learning & Memory, 14, 109-116.
https://doi.org/10.1101/lm.435807
Schwabe, L., Oitzl, M. S., Richter, S., & Schächinger, H. (2009). Modulation of spatial and
stimulus-response learning strategies by exogenous cortisol in healthy young women.

107
Psychoneuroendocrinology, 34, 358-366.
https://doi.org/10.1016/j.psyneuen.2008.09.018
Schwabe, L., & Wolf, O. T. (2012). Stress modulates the engagement of multiple memory
systems in classification learning. The Journal of Neuroscience, 32(32), 11042–11049.
https://doi.org/10.1523/JNEUROSCI.1484-12.2012
Schwabe, L., Tegenthoff, M., Höffken, O., & Wolf, O. T. (2013). Mineralocorticoid Receptor
Blockade Prevents Stress-Induced Modulation of Multiple Memory Systems in the
Human Brain. Biological Psychiatry, 74, 801-808.
https://doi.org/10.1016/j.biopsych.2013.06.001
Schwabe, L., & Wolf, O. T. (2013). Stress and multiple memory systems: From ‘thinking’ to
‘doing’. Trends in Cognitive Sciences, 17(2), 60-68.
https://doi.org/10.1016/j.tics.2012.12.001
Schwartz, D. H., Romans, S. E., Meiyappan, S., De Souza, M. J., & Einstein, G. (2012). The role
of ovarian steroid hormones in mood. Hormones and Behaviour, 62, 448-454.
https://doi.org/10.1016/j.yhbeh.2012.08.001
Shields, S. A., MacArthur, H. J., & McCormick, K. T. (2018). The gendering of emotion and the
psychology of women. In C. B. Travis, J. W. White, A. Rutherford, W. S. Williams, S. L.
Cook, & K. F. Wyche (Eds.), APA Handbook of the Psychology of Women: History,
theory, and battlegrounds (Vol. 1, pp. 189–206).
Spadaro, G., d’Elia, S. R. G., & Mosso, C. O. (2018). Menstrual knowledge and taboo TV
commercials: Effects on self-objectification among Italian and Swedish women. Sex
Roles, 78, 685–696. https://doi.org/10.1007/s11199-017-0825-0

108
Sommer, B. (1992). Cognitive Performance and the Menstrual Cycle. In J. T. E. Richardson
(Ed.), Cognition and the Menstrual Cycle (1st ed., pp. 39–66). Springer New York.
https://doi.org/10.1007/978-1-4613-9148-7_2
Steiner, M., Macdougall, M., & Brown, E. (2003). The premenstrual symptoms screening tool
(PSST) for clinicians. Archives of Women’s Mental Health, 6, 203-209.
https://doi.org/10.1007/s00737-003-0018-4
Sundström-Poromaa, I. (2018). The menstrual cycle influences emotion but has limited effect on
cognitive function. Vitamins and Hormones, 107, 349-376.
Takeda, T. (2023). Premenstrual disorders: Premenstrual syndrome and premenstrual dysphoric
disorder. Journal of Obstetrics and Gynaecology Research, 49(2), 510–518.
https://doi.org/10.1111/jog.15484
ter Horst, J. P., Kentrop, J., de Kloet, E. R., & Oitzl, M. S. (2013a). Stress and estrous cycle
affect strategy but not performance of female C57BL/6J mice. Behavioural Brain
Research, 241, 92–95. https://doi.org/10.1016/j.bbr.2012.11.040
ter Horst, J. P., Kentrop, J., Arp, M., Hubens, C. J., de Kloet, E. R., & Oitzl, M. S. (2013b).
Spatial learning of female mice: A role of the mineralocorticoid receptor during stress
and the estrous cycle. Frontiers in Behavioral Neuroscience, 7, 56.
https://doi.org/10.3389/fnbeh.2013.00056
Ussher, J. M. (2003). The ongoing silencing of women in families: An analysis and rethinking of
premenstrual syndrome and therapy. Journal of Family Therapy, 25(4), 388-405.
Ussher, J. M., Perz, J. & Mooney-Somers, J. (2007) The experience and positioning of affect in
the context of intersubjectivity: The case of premenstrual syndrome. International
Journal of Critical Psychology, 21, 144-165.

109
Vannuccini, S., Clemenza, S., Rossi, M., & Petraglia, F. (2022). Hormonal treatments for
endometriosis: The endocrine background. Reviews in Endocrine and Metabolic
Disorders, 23(3), 333-355.
van Ruitenbeek, P., Quaedflieg, C. W., Hernaus, D., Hartogsveld, B., & Smeets, T. (2021).
Dopaminergic and noradrenergic modulation of stress-induced alterations in brain
activation associated with goal-directed behaviour. Journal of Psychopharmacology,
35(12), 1449–1463. https://doi.org/10.1177/02698811211044679
Veselic, S., Jocham, G., Gausterer, C., Wagner, B., Ernhoefer-Reßler, M., Lanzenberger, R.,
Eisenegger, C., Lamm, C., & Losecaat Vermeer, A. (2021). A causal role of estradiol in
human reinforcement learning. Hormones and Behavior, 134, 105022.
https://doi.org/10.1016/j.yhbeh.2021.105022
Voon, V., Reiter, A., Sebold, M., & Groman, S. (2017). Model-Based Control in Dimensional
Psychiatry. Biological Psychiatry, 82(6), 391–400.
https://doi.org/10.1016/j.biopsych.2017.04.006
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures
of positive and negative affect: The PANAS scales. Journal of Personality and Social
Psychology, 54(6), 1063-1070. https://doi.org/10.1037//0022-3514.54.6.1063
Wirz, L., Reuter, M., Wacker, J., Felten, A., & Schwabe, L. (2017a). A haplotype associated
with enhanced mineralocorticoid receptor expression facilitates the stress-induced shift
from “cognitive” to “habit” learning. ENeuro, 4(6).
Wirz, L., Wacker, J., Felten, A., Reuter, M., & Schwabe, L. (2017b). A deletion variant of the
α2b-adrenoceptor modulates the stress-induced shift from “cognitive” to “habit” memory.

110
The Journal of Neuroscience, 37(8), 2149–2160.
https://doi.org/10.1523/JNEUROSCI.3507-16.2017
Wirz, L., Bogdanov, M., & Schwabe, L. (2018). Habits under stress: Mechanistic insights across
different types of learning. Current Opinion in Behavioural Sciences, 20, 9-16.
https://doi.org/10.1016/j.cobeha.2017.08.009
Wister, J. A., Stubbs, M. L., & Shipman, C. (2013). Mentioning Menstruation: A Stereotype
Threat that Diminishes Cognition? Sex Roles, 68, 19-31
Wittkuhn, L., Eppinger, B., Bartsch, L. M., Thurm, F., Korb, F. M., & Li, S.-C. (2018).
Repetitive transcranial magnetic stimulation over dorsolateral prefrontal cortex modulates
value-based learning during sequential decision-making. NeuroImage, 167, 384–395.
https://doi.org/10.1016/j.neuroimage.2017.11.057
Zerbes, G., Kausche, F. M., & Schwabe, L. (2020). Stress-induced cortisol modulates the control
of memory retrieval towards the dorsal striatum. European Journal of Neuroscience, 00,
1–15. https://doi.org/10.1111/ejn.14942
Zerbes, G. (2021). Stress and the nature of remembering: How stress and stress mediators alter
the contribution of multiple memory systems to retrieval [Dissertation]. Universität
Hamburg.

111
Appendix A: SurveyMonkey Physiological Questionnaire
1.
2.
3.
4.
5.

Have you felt muscle soreness in the last two days?
Have you had a headache in the last two days?
Have you felt more energized than usual in the last two days?
Did you have your menses/period in the last two days?
Did you get 7 to 9 hours of sleep per night in the last two days?

112
Appendix B: Demographic Questionnaire
1. Are you...?
1. Male
2. Female
3. Other, please specify: _________________________
2. Do you consider yourself to be trans (transgender, transsexual, or a person with a history
of transitioning sex)?
1. Yes
2. No
3. Don’t know
Answer questions 3 to 6 if you answered yes or don’t know
3. What was your assigned sex at birth?
1. Male
2. Female
3. Undetermined
4. What is your gender?
1. Male or primarily masculine
2. Female or primarily feminine
3. Both male and female
4. Neither male nor female
5. Don’t know
5. What gender do you currently live as in your day-to-day life?
1. Male
2. Female
3. Sometimes male, sometimes female
4. Third gender, or something other than male or female
6. What is your age?
1. 17-25
2. 25-30
3. 30-35
4. 35+
7. Please identify your ethnic identity: ____________________________
8. What year of university are you in?
1. First year
2. Second year
3. Third year
4. Fourth year
5. Fifth year or longer
9. What program/degree of study are you in? ________________________________
10. Is English your first language?
1. Yes
2. No

113
11. Are you right- or left-handed?
1. Right-handed
2. Left-handed
3. Both (ambidextrous)
12. Are you currently on any prescribed medication (this includes any hormonal medication
and contraceptives that secrete hormones)?
1. Yes
2. No
13. If yes, please describe medication: _____________________________
14. Have you taken any medication today (this includes any hormonal medication and
contraceptives that secrete hormones)?
1. Yes
2. No
15. If yes, please describe medication and when you last took it:
_________________________________________________________________

114
Appendix C: Menstrual Cycle Questionnaire
The following questions will ask about the characteristics of your menstrual cycle.
1. Are you currently menstruating?
a. Yes
b. No
2. On average, from month to month, would you consider your menstrual cycle regular?
a. Yes
b. No
3.
a. On average, how long is your menstrual cycle (in days)? This would include the
first day of your period to the last day before your next period. ________ OR
b. On average, what is the range of your menstrual cycle length (in days)? The
menstrual cycle length would include the first day of your period to the last day
before your next period. ________
4. To the best of your knowledge, when did your last menses/period begin?