The possible effect(s) of ovarian steroid hormones (i.e., estrogen and progesterone) on cognition and the brain have become of increasing interest to researchers over the last few decades. Most research in this domain has focused on sex hormone administration in animal models (e.g., Porcu et al., 2019), hormone replacement therapy (HRT) in peri- and postmenopausal females (e.g., Hogervorst et al., 2000, LeBlanc et al., 2001, Zhou et al., 2020), and fluctuations across the menstrual cycle (Le et al., 2020, Sundström-Poromaa and Gingnell, 2014, Sundström-Poromaa, 2018). Yet, the cognitive effects of oral contraceptive (OC) use remain understudied. OCs present a valuable opportunity to study the effects of sex hormones on cognition and they are an important target of study in-and-of-themselves, as they are one of the most common forms of hormonal contraception (used by more than 150 million individuals worldwide [United Nations, 2019]). OCs have been claimed to affect visuospatial ability, episodic memory, and executive control (Beltz et al., 2015, Gurvich et al., 2020; Peragine et al., 2019; Warren et al., 2014). However, notable limitations in the existing literature have prevented us from understanding the full scope of OCs’ effects on cognition (for reviews, see Beltz & Moser, 2019; Taylor et al., 2021, Warren et al., 2014). In the present study, we asked about the degree to which sex and OC use might influence visuospatial ability, episodic memory, and executive control. To improve upon the existing literature, we studied all three of these cognitive domains within the same participants and, for episodic memory and executive control, used composite neuropsychological scores, which provide more stable estimates of the underlying cognitive constructs being measured than single tests (Fortin and Caza, 2014, Glisky et al., 2022). We pre-registered our hypotheses, planned our sample size a priori using power analysis, and, to the best of our ability, controlled for several confounds related to OC use (i.e., hormonal formulation, timing).

The human brain is significantly influenced by the ovarian steroid hormones estrogen and progesterone. These hormones’ receptors are more highly expressed in females than males, and concentrated in several brain areas, including the prefrontal cortex (PFC) and hippocampus/medial temporal lobe (MTL). These two regions are primarily (although not exclusively) responsible for executive control and episodic memory, respectively. Estrogen exerts its effects on the brain via genomic (i.e., binding to nuclear receptors and altering gene transcription, leading to slower yet long-lasting effects) and non-genomic mechanisms (i.e., binding membrane receptors and activating signalling pathways, leading to more rapid effects) (see Barth et al., 2015; Luine et al., 2014). According to classical animal studies, the administration of estradiol (the most potent estrogen in pre-menopausal females) can exert neurotrophic effects on PFC and hippocampal structure by increasing synapse density and spine formation (Hasjan & Leranth, 2010; Leranth et al., 2008, McEwen and Alves, 1999, Tang et al., 2004, Woolley and McEwen, 1993). Further, estradiol treatment leads to enhanced memory in ovariectomized rats (Luine et al., 2003). In humans, estradiol promotes synaptic plasticity and neurogenesis throughout the brain (Azcoitia et al., 2011) and may modulate prefrontal dopamine function, which is heavily involved in memory and especially in executive control (Becker & Hu, 2008).

Given the positive neurobiological actions of estrogen and progesterone in the human brain, there has been notable interest in the field in how these hormones’ effects might contribute to behaviour and to individual differences in cognitive performance. Indeed, extensive research has shown that both endogenous (i.e., naturally occurring) and exogenous steroid hormones (i.e., in the form of OCs or HRT) can significantly influence a range of cognitive functions across the human lifespan and are key (although not exclusive; see Discussion) determinants of cognitive sex differences (for reviews, see Boss et al., 2013; Gurvich et al., 2018, Hamson et al., 2016; Hara et al., 2015; Khadilkar and Patil, 2019, Kimura, 1996; Nunez, Maraver, & Colzato, 2020).

Given their structural similarity to endogenous sex hormones, the synthetic hormones contained in OCs may cause significant changes in cognition and the brain. Modern combined OCs contain ethinyl estradiol and a progestin, which are synthetic forms of estradiol and progesterone, respectively. These hormones provide negative feedback to the hypothalamic-pituitary-gonadal (HPG) axis, leading to a downregulation of the endogenous estradiol and progesterone availability in OC users, thus inhibiting ovulation (Rivera et al., 1999). Most combined OCs reduce endogenous estradiol and progesterone to levels equivalent to those in the early follicular phase (i.e., menses) of the menstrual cycle, when estradiol and progesterone are at their lowest (Sahlberg et al., 1987). Both the suppression of endogenous hormones and the addition of synthetic hormones due to OCs may have neurobiological consequences, but these are difficult to dissociate and properly characterize without the proper methods, such as hormone assays (Hampson 2020; Taylor et al., 2021), which many studies so far have lacked. Furthermore, combined OCs also differ in terms of their progestin component: 1st and 2nd generation OCs contain androgenic progestins (i.e., norethindrone, levonorgestrel) that are derived from testosterone and have equivalent side-effect profiles (Batur et al., 2003, De Leo et al., 2016, Guerra et al., 2013). 3rd generation OCs, containing desogestrel or norgestimate, on the other hand, exert less androgenic actions (Kuhl, 1996), whereas 4th generation OCs, containing drospirenone or dienogest, are anti-androgenic and bind to progesterone receptors (Sitruk-Ware and Nath, 2010, Wiegratz and Kuhl, 2006). Accumulating evidence suggests that, depending on their progestin component, OCs may exert either feminizing or masculinizing effects on sex-typed cognitive functions, including visuospatial ability (favouring males) and verbal memory (favouring females) (Pletzer & Kirschbaum, 2014).

One of the most prominent reported sex differences favouring males is in mental rotation (Shepard & Metzler, 1971), a visuospatial ability which involves forming a mental representation of an object and recognizing it from a different orientation in space (Linn and Petersen, 1985, Voyer et al., 1995). The average reported sex difference in young adults is Cohen’s d = 0.37, with some studies finding effect sizes as large as d = 0.9 (see Linn and Petersen, 1985, Voyer et al., 1995). The level of androgenic activity of the progestin component of the OC formulation, and not overall OC use, may contribute to different mental rotation performance and amplify the sex difference observed on this task (Peragine et al., 2019; Wharton et al., 2008): According to these studies, females taking older generation (i.e., androgenic) OCs have enhanced mental visuospatial abilities and perform similarly to males on tasks of mental rotation, whereas anti-androgenic OC users and non-users have poorer performance. Correspondingly, higher circulating estrogen may also contribute to worse spatial ability (Beltz et al., 2015, Mennenga et al., 2015), whereas lower circulating estrogen may have the opposite effect (Peragine et al., 2019). However, a more recent, high-powered study of 528 naturally-cycling females found that fluctuations in estradiol did not significantly predict mental rotation (Shirazi et al., 2021). Although numerous studies have found no significant mental rotation differences between OC users and non-users (Gordon and Lee, 1993, Griksiene et al., 2018, Islam et al., 2008, Mordecai et al., 2008) – and one study even reported a negative effect of OC use (Griksiene & Ruksenas, 2011) – many of these have included heterogeneous OC groups and did not categorize their participants based on OC formulation or progestin type (Warren et al., 2014). Taken together, these findings suggest that researchers should consider the generation of OCs taken by participants as well as their levels of circulating estradiol as these may be important predictors of their mental rotation performance.

Females show an advantage over males in episodic memory (i.e., one’s long-term, explicit recollection of past personal experiences), especially for verbal materials (Andreano and Cahill, 2009, Asperholm et al., 2019). This sex difference may be due in part to the beneficial effects of ovarian steroid hormones such as estradiol, which has been argued to enhance explicit memory (Frick, 2015, Luine and Frankfurt, 2013, Phillips and Sherwin, 1992). Similarly, the addition of exogenous steroid hormones through OC use may also modulate verbal memory, with one study finding that OC users perform similar to naturally cycling females in phases characterized by high estradiol and progesterone (Gogos, 2013). Mordecai et al. (2008), on the other hand, found no significant differences between OC users and non-users on a cognitive battery assessing verbal and visual memory, mental rotation, and attention. Yet, OC users who were tested during the active pill phase performed better than users in the inactive pill phase of their cycle. Additionally, time of pill ingestion may affect verbal memory: OC users who ingested their pill close to two hours prior to cognitive testing (when plasma estradiol levels are estimated to be at their highest) had better retention on a story recall task than those who took their pill afterward (Peragine et al., 2019). These findings contrast with those of Islam et al. (2008), who found no significant differences in verbal memory (or any other tasks in their cognitive battery) between OC users and non-users, nor across pill cycle phases. Other studies have reported null findings in this cognitive domain (Gordon and Lee, 1993, Mihalik et al., 2009, Wharton et al., 2008); however, these studies used rather heterogeneous OC user groups with respect to pill phase and ingestion time.

It follows that OC use might positively affect verbal memory in periods when circulating estradiol is at its highest (e.g., during active pill phase, 1-2h following pill ingestion). Finally, according to several studies, progestin type does not appear to affect verbal memory (Gogos, 2013, Gurvich et al., 2020, Wharton et al., 2008).

More controversial than the cases of mental rotation and episodic memory is that of executive control. Executive control – highly related to the concepts of executive function, cognitive control, and attentional control – may preferentially recruit the frontal lobes (Alvarez & Emory, 2006; McCabe et al., 2010). In human neuroimaging research, males and females may differ in the neural networks recruited by executive control tasks (Gaillard et al., 2021), which may be related to differences in levels of circulating ovarian steroid hormones. For instance, among naturally cycling females, performance on frontal lobe-dependent cognitive tasks, such as those measuring inhibitory control and working memory, may improve during periods of high estradiol levels (i.e., late-follicular phase of the menstrual cycle) (Colzato et al., 2010; Jacobs and D'Esposito, 2011, Rosenberg and Park, 2002; for a review, see Sundström-Poromaa & Gingnell, 2014).

Several neuroimaging studies have also suggested links between OC use and structural and functional changes in the frontal lobes and other regions belonging to the executive control network (Gingnell et al., 2016, Kirschbaum et al., 1999, Petersen et al., 2014, Pletzer et al., 2010, Pletzer et al., 2014, Pletzer et al., 2015, Pletzer et al., 2016, Sharma et al., 2020). When considering progestin type, individuals using OCs with androgenic progestins had smaller grey matter volume in the PFC than naturally cycling females, which could imply that androgenic progestins may exert detrimental effects on frontal lobe-related cognition (i.e., executive control) (Pletzer et al., 2015). Anti-androgenic OC users also showed decreased resting-state functional connectivity in the medial PFC compared to non-users (Pletzer et al., 2016). Furthermore, when considering OC pill phase, compared to naturally cycling females, combined OC users in the active pill phase had decreased cortical thickness in the lateral orbitofrontal cortex, a region involved in executive control (Petersen et al., 2015). It therefore appears that both the androgenic activity of progestins contained in OCs as well as the timing of hormone administration may influence the effects of OC use on brain structure and function.

Yet, from a cognitive perspective, unlike verbal episodic memory and visuospatial ability, there appear to be no consistent sex differences in executive control (Gaillard et al., 2020; Grissom & Reyes, 2019). Regarding OCs, one study reported that OC use worsened verbal fluency, with even worse performance among androgenic than anti-androgenic OC users (Griksiene & Ruksenas, 2011). However, another study found that OC use had no significant effects on verbal fluency (Mordecai et al., 2008). In fact, most cognitive studies of executive control have found little support for OC-related effects (Gogos, 2013, Gurvich et al., 2020, Islam et al., 2008, Wright and Badia, 1999).

Overall, there appears to be a dissociation between the effects of biological sex and OCs on executive control-related brain areas/networks and their contributions to executive control performance. Perhaps the executive control tests used in previous studies did not accurately capture the influence of biological sex or OCs or, alternatively, the effects of biological sex and OCs are captured by neuroimaging measures but were too subtle to appear as changes in performance. We were therefore interested in investigating this relationship further by using more robust measures of executive control in a larger sample of participants than previous studies.

Based on the small existing literature outlined above, we are beginning to understand the effects of OC use on cognition; however, this picture is not yet clear due to several limitations of previous studies. First, most studies were underpowered to detect a significant effect of OCs on cognition (Warren et al., 2014). Second, many studies used single cognitive tests which were designed for clinical use in cognitively impaired individuals and may have not been sensitive enough to assess subtle changes in cognition in healthy participants (Warren et al., 2014). Finally – although recently researchers have been more attuned to the multidimensionality of OCs – previous work often compared heterogeneous groups of “current OC users” to “non-users”, thus overlooking a number of moderating factors that may differentially affect cognition, such as hormone formulation (i.e., combined versus progestin-only, generation/androgenic activity), OC regimen (i.e., monophasic versus multiphasic), age at initiation, OC use history, and duration of use (Hampson, 2020; Taylor et al., 2021). Addressing these shortcomings of previous work is necessary to establish a more solid understanding of OC effects on cognition.

The purpose of the present cross-sectional, quasi-experimental study was to investigate how biological sex and OC use might affect individual mental rotation, episodic memory, and executive control in young adults, by addressing several of the limitations in the extant literature. By using a more appropriate study design (e.g., including a larger N with a priori power analysis, considering progestin androgenicity) and sets of cognitive tests (e.g., using composite scores for episodic memory and executive control instead of single cognitive tasks), this study should provide more insight into better understanding the relations between OC use and cognition.

We used a cognitive test battery inspired by Glisky and colleagues’ two-factor battery thought to measure episodic memory and executive control (Davidson and Jensen, 2022, Glisky et al., 1995, Glisky et al., 2001, Glisky et al., 2022, Glisky and Kong, 2008, Young-Bernier et al., 2012). Based on factor analysis results, for each participant we derived two composite z-scores, one representing relative episodic memory and the other executive control. Using composite scores presents several psychometric advantages over using single neuropsychological tests, by reducing the total number of statistical comparisons and providing a more robust estimate of the underlying cognitive constructs. To the best of our knowledge, the Glisky (or Glisky-inspired) composite scores have not yet been used to investigate sex differences in young adults, nor to study OC use in such populations. Additionally, we aimed to examine the effects of other OC-related factors on mental rotation, episodic memory, and executive control which may affect cognition, such as pill phase, time of OC pill ingestion, and duration of OC use.

We hypothesized:

1. Depending on the androgenic activity of OCs used in our sample, males would outperform naturally cycling females but not OC users on mental rotation (Linn & Petersen, 1985; Peragine et al., 2019; Voyer et al., 1995, Wharton et al., 2008).

2. Given the growing literature on the beneficial effects of estradiol on memory function (Frick, 2015, Luine and Frankfurt, 2013, Phillips and Sherwin, 1992), females (both naturally cycling and OC users) would perform similarly to one another and each outperform males on the memory composite.

3. a) In accord with the literature showing no clear sex difference in executive control, male and female groups would perform similarly on the executive control composite (Gaillard et al., 2020; Grissom & Reyes, 2019).

b) Several previous neuroimaging studies have argued that fluctuations in endogenous ovarian steroid hormones (estradiol, progesterone) and exogenous, synthetic versions of these hormones (ethinyl estradiol, progestins) may positively affect brain areas and networks supporting executive control, and thus that female biological sex and OC use (specifically the progestin component and the availability of exogenous hormones) should each contribute to better executive control (Kirschbaum et al., 1999, Petersen et al., 2014, Pletzer et al., 2010, Pletzer et al., 2014, Pletzer et al., 2016). However, cognitive studies have shown few convincing positive effects of OC use (Gurvich et al., 2020, Gogos, 2013, Islam et al., 2008, Mordecai et al., 2008, Wright and Badia, 1999). We therefore were curious but skeptical about OC users showing better executive control composite scores than naturally cycling females.

4 Longer duration of OC use might be associated with better executive control and memory (Egan and Gleason, 2012, Karim et al., 2016).