Master’s Thesis, 15 ECTS
Master’s Program in Cognitive Science, 60 ECTS Spring term 2018
Supervisors: Johan Eriksson, Tiziana Pedale
GENDER DIFFERENCES IN
UNCONSCIOUS VISUAL
WORKING MEMORY
Thanks to my supervisors Johan Eriksson and Tiziana Pedale, who not only have given me invaluable insight into the theoretical world of academics, but have also given me the opportunity to catch a glimpse of what it is like to practically be involved in conducting research.
Thanks to my friends and highly esteemed proofreaders, Dinah van Bavel and Maaike Oosterling, for always so selflessly finding a way through my rambling, both in- and outside of the boundaries of this thesis. I sincerely apologize for my lack of ability to use any kind of punctuation properly.
Abstract
Recent research has shown that working memory tasks can be performed with information that hasn’t been consciously perceived. This provides new opportunities in research concerning possible limitations and influential factors on unconscious working memory. Gender has been demonstrated to be a factor affecting conscious visual working memory tasks and could likewise influence unconscious visual working memory. An analysis of behavioral data, obtained in three similar unconscious visual working memory tasks (n = 89), was performed. Three ANCOVAs were conducted to establish whether there was a significant effect of gender on unconscious working memory accuracy, response time (RT) and speed-accuracy tradeoff (SAT) across the three datasets. The analysis demonstrated a significant advantage in response time for female participants compared to male participants. Implications of this observation, such as male and female response strategies and possible social implications of unconscious processes, are discussed in this thesis.
Keywords: working memory, consciousness, gender, visuospatial, response time Abstrakt
Nyligen har arbetsminnet bevisats fungera omedvetet. Detta öppnar nya dörrar för forskning om möjliga begränsningar och påverkande faktorer på omedvetet arbetsminne. Kön, som har betydande inflytande på medvetet visuellt arbetsminne, kan på samma sätt också påverka det omedvetna visuella arbetsminnet. En analys av data, insamlad vid tre liknande visuella uppgifter där det omedvetna arbetsminnet användes (n = 91), utfördes. ANCOVAs utfördes för att fastställa huruvida det fanns betydande effekter utifrån kön på det omedvetna arbetsminnets noggrannhet, svarstid samt avvägning mellan snabbhet och noggrannhet, oavsett uppgift. Analysen visade ett signifikant övertag vad gäller kvinnors svarstid i jämförelse med manliga deltagare. Detta övertag gäller generell svarstid och en fördelaktig avvägning mellan snabbhet och noggrannhet. Konsekvensen av dessa observationer, såsom olika svarstaktik för män och kvinnor, möjlig samhällspåverkan av omedvetna processer, diskuteras i denna uppsats.
Gender differences in unconscious visual working memory
In his book “The Magical Number Seven, Plus or Minus Two”, George Miller (1956) discussed the limitations and possibilities of a then yet to be named ‘Working Memory’. He argued that while one performs a task, a certain structure or process in the brain allows one to temporarily hold and manipulate information. It is now clear that working memory is not just one structure or process, but relates to several intellectual abilities. Working memory underlies other functions related to cognitive control and executive functioning: it can guide behavior, interact with long-term memory, update information, and aid in executing a plan (Nyberg & Eriksson, 2016). Due to the complicated and integrative nature of working memory, some argued that such a mechanism had to hold a close relation to consciousness1 (James, 1890). Others even assumed that consciousness was a necessary requirement for working memory to operate (Richet, 1886) or that we are conscious of our working memory content and processes (Atkinson & Shiffrin, 1971).
Recent studies have demonstrated that one can perform working memory tasks without conscious awareness. In a review paper, Soto and Silvanto (2014) argued through a number of studies for a new way of dissociating conscious awareness from working memory processes. How participants perceived their working memory content through for example confidence ratings, could differ from the actual working memory content or accuracy (Bona, Cattaneo, Vecchi, Soto, & Silvanto, 2013; Bona & Silvanto, 2014; Magnussen, 2000). This refutes Atkinson’s claim that we are conscious of our own working memory content and processes. Consciousness is likewise not a necessary requirement for working memory to operate, as Soto and colleagues (2011) demonstrated; participants performed significantly above chance in a visual unconscious working memory task, even when a target was accompanied by a distractor. Unconscious working memory is a relatively complex and integrative mechanism. Its limitations and operational mechanisms are unclear, and so are the factors that could possibly influence unconscious working memory. A logical step in the process of discovering influential factors on unconscious working memory would be to uncover if factors that influence conscious working memory could have an effect on unconscious working memory.
One factor that has been shown to significantly influence different aspects of working memory is gender. Gender2 significantly influences different aspects of working memory, but there is not one gender that is more skillful in all working memory operations. Due to the multi-faceted nature of working memory, this comes as no surprise. Goldstein and colleagues (2005) established an auditory working memory difference in functional imaging studies, where female participants exhibited greater Blood Oxygen Level Dependent (BOLD) signal changes in the neural correlates of working memory – the prefrontal cortex – than male participants in auditory verbal working memory tasks. Voyer et al. (2017) performed a meta-analysis on visual-spatial working memory and argued that males demonstrated an advantage in this domain, but only for specific tasks. For example, mental rotation tasks were consistently performed better by males (Maeda & Yoon, 2013), whereas an earlier study by Voyer demonstrated a female advantage for location memory (Voyer et al., 2007). Duff and Hampson (2001) likewise demonstrated a female advantage in a spatial working memory task that relied heavily on location memory. Thus, an advantage for either gender on conscious working memory seems to be task-specific, but there appears to be an indication that women
1 Consciousness will in this thesis be defined as “the state of being aware of one’s surroundings” due to the nature of the experimental task and the broad implementation this definition has when consulting literature regarding working memory and consciousness.
perform better on location memory tasks. It is however unknown whether gender has any effect on unconscious visual working memory. It is therefore valuable to research whether gender affects performance on an unconscious visual working memory task, such as the tasks analyzed in this thesis. Comparisons such as the influence of gender on unconscious working memory have not been performed previously due to the relatively young nature of unconscious working memory research. This thesis aims to take a first step in exploring if gender affects unconscious working memory.
In order to investigate if gender has any effect on unconscious visual working memory, this thesis uses data collected at the Umeå center for Functional Brain Imaging (UFBI). The research consists of three variants of a task. The task requires participants to retain the location of a symbol, often without having consciously perceived it. Participants performed a series of trials where they had to decide whether a location shown at the end of the trial corresponded to the location of the symbol. Continuous Flash Suppression (CFS) ensured that stimuli could be rendered unconscious. By presenting a stimulus in low contrast to one eye, while colored images of random composition (so-called “Mondrians”) are flashed to the dominant eye, experience of the stimulus could be suppressed. Figure 1 visually represents how stimuli were suppressed using CFS in the task that is analyzed in this thesis.
Performance was measured through accuracy, response time (RT) and speed-accuracy tradeoff (SAT). SAT was operationalized by comparing the RT corrected for errors made between the genders. The tasks performed are visualized in Appendix A and elaborated upon in the method section. With these tasks in mind, the research question this thesis aims to answer is as follows: Does gender influence accuracy, response time and/or speed-accuracy tradeoff in a visual unconscious object location working memory task?
Because the task analyzed in this thesis is an object location memorization task, female participants might demonstrate a slight advantage in accuracy, RT or SAT. With the academic literature in mind, the null hypothesis would be that gender has no significant effect on accuracy, RT or SAT in an unconscious visual working memory task. The alternative hypothesis argues the opposite; gender will have an effect on one or more of these task outcomes, and this effect will be characterized by a female advantage.
Method Participants
Data used for analyses was obtained in three experiments. In total, 91 data points were obtained (n1 = 28, 13 female (f), mean age = 27.9, SD = 7.07; n2 = 32, 17f, mean age = 26.8,
SD = 4.53; n3 = 31, 19f, mean age = 27.6, SD = 4.60). No other demographics but age and
gender were collected. Data from two participants was excluded from analysis because they participated in two of the three studies. The total analyzed dataset contained 89 independent participants. Participants were recruited via online advertisements and posters on the campus of Umeå University. Participants were almost all right-eye dominant and right-handed, and had normal or corrected to normal vision. In the first experiment, six (6) participants were left-eye dominant, of which two (2) were left-handed. Two (2) participants were left-handed, but right-eye dominant. For left-eye dominant participants, stereoscope setup as discussed below was inverted. Participants were financially compensated for their time. All participants gave informed consent to be included in the study. All procedures performed in experiments involving human participants were in accordance with the ethical standards of and approved by the regional ethical committee. The three studies were conducted in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Instruments and Materials
The task was carried out with the use of a stereoscope. A stereoscope allows two separate images to be perceived as one. Without a stereoscope, a person usually sees one object that is perceived independently by both eyes. The brain converges these two
perceptions into a coherent image. A stereoscope uses mirrors to present two images that can differ from each other, to both eyes. The brain will still converge both images as if perceiving one and the same image. CFS was used to render stimuli unconscious. E-prime was the programme used to run the task.
Procedure
All three experiments required the participants to perform a working memory task. The first and third experiment took place in an MRI, while the second experiment did not. All experiments varied slightly, but shared their major features. All tasks were a visual object location memorization task where the participant was asked to retain the location of a certain symbol in a series of delayed match-to-sample trials. It was tested whether or not the
participant could discriminate a target from a distractor in an unconscious setting.
Through the stereoscope, this symbol could be presented consciously or
unconsciously to the participant. If experience of the stimulus was suppressed, the trial was labeled ‘unconscious’. The trial was labeled ‘conscious’ if the stimulus was presented to the dominant eye. In some trials the stimulus was not present. The trial was then labeled ‘absent’. See Appendix A for a full overview of the tasks and trials. When performing the task,
participants were not explicitly requested to give their response as quickly as possible, nor were they informed that their response time was measured. They were however informed that they had a window of five seconds to give an answer.
Experiment 1. Data was collected over 3 sessions and 222 trials in total. See table A1 for a breakdown of trials (Appendix A). The task was a series of delayed match-to-sample trials where the cue was always presented consciously and the sample was presented
consciously, unconsciously, or it was absent. The cue was either a circle or a diamond symbol (see figure 2) and indicated the relevant shape (the target) that should be remembered. After the target, there was a small delay; a grey screen was presented for 250ms. The delay was followed by the sample, as illustrated in figure 1. The sample consisted of the simultaneous presence of the target and the distractor. Both occupied one of the four corners, leaving two corners unoccupied. The participant then held the location of the target in mind during a 5 – 15s delay. The location was always one of the four corners in the image. After the delay, the participant was asked if the probe (see figure 3) presented pointed towards the location of the target.
The probe could either point towards the target (‘match’ condition), the distractor (‘distractor match’ condition) or to a location where there was no target nor distractor (‘nonmatch’ condition). If the participant responded ‘yes’ if the probe pointed towards the distractor, this was reported as a ‘False Alarm’ (FA).
Finally, at the end of each trial, a subjective measure of conscious experience of the participant was acquired via the Perceptual Awareness Scale (PAS). The participant was asked to report whether they saw nothing, saw something (e.g. a flash of a grey symbol or an outline), or saw clearly. If the participant reported seeing ´something´ three times or more per ten trials, the contrast between the symbols and background displayed to the non-dominant
Figure 2. The ‘cue’ - the shape participants had to keep in mind. It was always presented consciously.
eye (see figure 1) was automatically decreased. Diminished contrast combined with CFS ensured that the symbol did not reach the threshold of conscious perception and that every participant had a contrast that was optimized to their threshold. If the participant always reported seeing ‘nothing’, the contrast would be increased so the participant was still able to unconsciously perceive the symbol. See figure A1 (Appendix A) for a visual representation.
Experiment 2. Data was collected over four sessions and a total of 312 trials. See table A2 for a breakdown of trials (Appendix A). The task was altered, as the ‘nonmatch’ condition in unconscious trials was not included. The probe could thus only point towards the correct symbol or the distractor. By doing so, more trials could be devoted to comparing accuracy between the ‘match’ and the ‘distractor match’, which increased the statistical power of this comparison. See figure A2 for a visual representation.
Experiment 3. Data was collected over 3 sessions and a total of 222 trials. See table A3 for a breakdown of trials (Appendix A). The ‘nonmatch’ condition in unconscious trials was excluded from this experiment as well. The trials in this experiment additionally included conscious and unconscious samples where only a target was present, no distractor. Trials that only contained a target were not included in the analysis. See figure A3 for a visual
representation. Analysis
In order to answer the question whether gender was of any effect on the unconscious working memory performance, accuracy, RT, and SAT of participants in the task were compared on the basis of the gender of the participant. An average performance over all unconscious trials per participants was gathered and is used in the current analysis. Only unconscious or absent trial conditions are analyzed. To calculate the SAT, the inverse efficiency score (IES) equation was used. To ensure that this analysis focuses on accuracy, it is necessary to confirm that the participants did perform relatively accurately to the task at hand, i.e. they perform above chance. To do so, a one-sample t-test is performed, that tests the performance measure versus the value of 0. The performance measure used at UFBI is the performance measure used in this thesis and is as follows:
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 ℎ𝑖𝑡𝑠 − 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑙𝑠𝑒 𝑎𝑙𝑎𝑟𝑚𝑠
The number of false alarms indicates times where the probe pointed to the position of the distractor, and the participant responded ‘yes’ to the question whether or not the probe pointed towards the target. This performance measure measures the ability to differentiate between target and distractor in an unconscious trial.
Accuracy was operationalized similarly on a scale from 0 to 1 in an unconscious trial. Conscious trials were not included in the analysis. How often one correctly responds to the probe is calculated by subtracting the proportion of FAs from the proportion of total ‘hits’. These ‘hits’ represented all the times the participant responded ‘yes’ to the probe, which could point either to the distractor or the target. Data analysis excluded nonmatch trials as this condition was only an unconscious option in the first experiment. RT was operationalized by average response time to each probe in milliseconds (ms). Gender and target side were variables of interest. Ongoing research at the UFBI indicates that in the task researched in this thesis, a lateralization effect is present. When the target was presented in the left visual hemifield, performance in the unconscious working memory trials improved (see Appendix A; table A1, A2 and A3 for breakdown of trials per target side). Therefore, target side will be considered as part of the analysis to investigate the effect gender might have in interaction with target side, in order to maximize the chance of revealing any potential gender effects.
A repeated measures General Linear Model was created, where the dependent variable was divided per target side and gender was always the independent variable. One covariate was the type of experiment the participant was part of, as the slight differences between experiments might affect performance. In the ANCOVA concerned with RT, reaction time on absent trials was included as a covariate as ‘Baseline RT’. This was done in order to exclude the option that participants of one specific gender are solely faster at decision making tasks and to ensure that the variable of interest is still concerned with the unconscious working memory task.
In order to control whether shorter RTs occurred at the expense of accuracy, SAT was calculated using the inverse efficiency score (IES) (Townsend & Ashby, 2014), which is defined as
𝐼𝐸𝑆 = 𝑅𝑇 1 − 𝑃𝐸
Where RT is the response time for a correct answer of a participant and PE is the proportion of errors in the task. The higher PE or the higher RT is, the larger the IES and the weaker the SAT. PE can be defined in two ways: by the correct probe positions that one responded ‘no’ to, or by the False Alarm probe positions that one responded ‘yes’ to. These types of PE’s can respectively be called PEmiss and PEFA. PEmiss is operationalized by subtracting correct
responses as defined for operationalizing ‘accuracy’ from 1, and is used to calculate IESmiss.
PEFA is defined by the proportion of FA responses to the probe and is used to calculate IESFA.
Each participant thus has an IESmiss and an IESFA that can be averaged out to a personal IESµ.
The formula for IESµ would have as definition
𝐼𝐸𝑆µ =
𝐼𝐸𝑆 + 𝐼𝐸𝑆
2
For all participants from one gender, IESµ values can also be summed and subsequently
averaged out to IESµf for female participants or IESµm for male participants. The formula for
the IESµf would be defined as
𝐼𝐸𝑆µ = ∑ 𝐼𝐸𝑆µ 𝑛
Where nf is the number of female participants. IESµ is analyzed in an ANCOVA to evaluate
if there are any significant gender differences. Results
All assumptions – no outliers, normal distribution, sphericity, homogeneity - were met for all ANCOVAs. The research question revolved around the effect of gender on
unconscious working memory accuracy, RT, and SAT. Target side and type of experiment could be of possible influence on this outcome, with a possibility of interaction between target side and gender. Baseline RT was used as a covariate in the ANCOVA that aimed to explore any possible effects of gender on RT.
Average Accuracy
demonstrated that accuracy (M = 0.058, SD = 0.12 was significantly higher than 0, t(88) = 4.460, p < .001.
Accuracy
Accuracy was examined with a 2 × 2 × 3 (Gender [male, female] × Target side [left, right] × Experiment [1,2,3]) mixed ANCOVA. The repeated measures ANCOVA concerned with unconscious working memory accuracy demonstrated no significant effects of gender, target side, or experimental task. There was also no significant interaction between gender and target side. See Table 1 for within-subject effects and Table 2 for between-subjects effects.
Response Time
RT was examined with a 2 × 2× 3 (Gender [male, female] × Target side [left, right] × Experiment [1,2,3]) mixed ANCOVA with Baseline RT as an additional covariate. The repeated measures ANCOVA concerned with unconscious working memory RT revealed a significant effect of gender on unconscious working memory RT, F(1,85) = 4.021, p = .048, ηp2 = .045. Baseline RT had a significant effect on unconscious working memory RT, F(1,85) = 268.66, p = .000, ηp2 = .760 (see Table 3). There was no significant interaction with target side (see Table 4). A pairwise comparison demonstrated that the effect was characterized by a female advantage. There is significant difference in unconscious working memory RT for females versus males (p = .048), where females responded 89 ms faster than males (see Table 5).
Table 1
Analysis of covariance to reveal differences in working memory accuracy per target side and relevant interactions (N = 89)
Source Type III Sum
of Squares df Mean Square F Sig. Partial Eta Squared Target Side .057 1 .057 1.385 .242 .016 Target Side*Experiment .004 1 .004 .091 .764 .001 Target Side*Gender .006 1 .006 .143 .764 .002 Error 3.558 86 .041 Table 2
Analysis of covariance to reveal differences in working memory accuracy per gender and experiment (N = 89)
Source Type III Sum of Squares df Mean Square F Sig. Partial Eta Squared Experiment .025 1 .025 .896 .347 .010 Gender .005 1 .005 .173 .678 .002 Error 2.447 86 .028 Table 3
Analysis of covariance to reveal differences in working memory RT per gender and target side (N = 89)
Source Type III Sum of Squares df Mean
*p < .05
Speed-Accuracy Tradeoff
SAT was defined the RT that was corrected for errors via the IES calculation by Townsend and Ashby (2014) (see Table 6). Female participants received a lower score on the IES than men, resulting in a more beneficial SAT. Their faster RT does thus not seem to
Squared Experiment 241.120 1 241.120 .003 .957 .000 Baseline RT 21802965.11 1 21802965.11 268.663 .000* .760 Gender 326292.196 1 326292.196 4.021 .048* .045 Error 6898054.448 85 81153.582 Table 4
Analysis of covariance to reveal differences in working memory RT per target side and relevant interactions (N = 89)
Source Type III Sum
of Squares df Mean Square F Sig. Partial Eta Squared Target Side 1251.923 1 1251.923 .143 .707 .002 Target Side*Experiment 16249.873 1 16249.873 1.851 .177 .021 Target Side*Baseline RT 2321.885 1 2321.885 .264 .608 .003 Target Side*Gender 11323.425 1 11323.425 1.290 .259 .015 Error 746222.554 85 8779.089 Table 5
Pairwise comparisons in analysis of covariance to reveal
differences in working memory RT on gender (N = 89) 95% Interval for Difference Confidence (I) Gender (J) Gender Mean Difference (I-J)
Std. Error Sig. Lower Bound
Upper Bound
f m -88.885 44.328 .048* -177.020 -.749
*p < .05
negatively affect accuracy. SAT was examined with a 2 × 3 (Gender [male, female] × Experiment [1,2,3]) ANCOVA with Baseline RT as an additional covariate. The ANCOVA revealed no significant results (see Table 7).
Table 6
IES per gender (N = 89)
Discussion
In sum, significant gender differences were found in RT, but not in SAT and accurate performance. The null hypothesis can therefore be rejected. The strong significant effect of baseline RT suggests that women might simply be better at decision-making tasks. However, as SAT was in favor of female participants, it is not unthinkable that the results yielded in this analysis might indicate a task-specific female advantage. Other studies that focused on
reaction time to a task similar to the current task, found that men are overall faster than women (Der & Deary, 2006; Jain et al., 2015; Misra, Mahajan, & Maini, 1985). These studies requested participants to react to one stimulus in a way that is similar to what is required of participants when they see the probe. The question is then: what could explain the seemingly contradictory results of this thesis? A key difference between RT studies such as Der and Deary’s study and the current study is whether participants were instructed to respond as fast as possible. Contrary to previously mentioned studies, this task did not instruct participants to react as soon as possible. It solely asked participants to respond ‘yes’ or ‘no’ to the displayed probe within a five-second time window. This distinction between ‘reacting’ and
‘responding’ is of vital importance. Reacting is a motor function, while responding requires executive functioning. A reflex can be a reaction, while an answer to a question requires a response. Men consistently score better on reaction time studies, an advantage attributed to a stronger motor control system (Jain et al., 2015). However, this effect is more variable and less significant when a response is required (Der & Deary, 2006). The male advantage is also minimized by withholding the information that response time will be measured, as this diminishes the element of competitive behavior. Men are more competitive than women in physical and attentive tasks (Cashdan, 1998), causing shortened reaction time (Wong et al., 2017).
This minimized male advantage does however not yet clarify the female advantage in both RT and SAT. An explanation for this advantage might be that when making a choice on a screen, men and women adopt different strategies. Women are more likely to employ a
Average IES Male Participants Female Participants
IESmiss 3758.758 2713.055
IESfa 2493.324 2105.011
IESµ 3126.041 2409.033
Table 7
Analysis of covariance to reveal differences in working memory SAT based on IES per gender (N = 89)
Source Type III Sum of Squares df Mean
Square F Sig. Partial Eta
Squared
Experiment .005 1 .005 1.240 .269 .014
Baseline RT 2.881E-6 1 2.881E-6 .001 .978 .000
Gender .002 1 .002 .618 .434 .007
serial ‘left-to-right’ strategy as if they are reading, while men are more likely to
‘dichotomize’ – i.e. imagine splitting the screen from the middle and gather data from nearby stimuli from there (Adam et al., 1999). The starting point for scanning a screen for a stimulus for women would then be top left, followed by a motion that is similar to a reading eye gaze, whereas men would start in the center and search for a stimulus from there. These different strategies could possibly explain the advantage women have, as the probe would always point towards one of the corners and never to the middle. Men would therefore always have a slight time penalty. However, if women tend to scan a screen as if they are reading, the bottom right corner might also be accompanied by a penalty. The extent of the penalty for men would therefore determine whether the scanning strategy is a relevant factor in gender differences in RT. Although there was no significant interaction between gender and target side, perhaps there would be a significant effect of the top left corner as a starting point on this female advantage in RT Ongoing research at UFBI has indicated that the side of the target affects unconscious accuracy, so it would be interesting to see if accuracy, response time and SAT would improve if the probe would signal towards the top left corner. If spatial orientation affects reaction time in this task, maybe it could affect unconscious decision making altogether.
Undoubtedly, a theory on spatial navigation such as Adam’s (1999) is not argued to be the sole explanation of differences in RT and SAT. Both significant and non-significant results in this thesis can be confounded by several factors. Gender is a complicated variable that is strongly mediated by socialization and expectation.Correlating gender with
unconscious working memory performance is not only obstructed by the fact that
performance is task-dependent. Although the variable ‘gender’ is in this thesis treated as a binary variable that relied on the researcher’s assumption of a participants gender based on their appearance, the concept of gender is not this straightforward. The gender of participants can cause different performances due to the socialization that is often tied to that gender. Serbin and colleagues (1990) provided a model that describes the role of socialization to frequently be a mediator between gender and cognitive functioning based on their review of several studies in the field. They demonstrated that, for example, visuospatial skills do not differ across genders. At a young age girls experience a ‘practice deficit’ in tasks that concern hand-eye coordination or quick visuospatial performance due to them being assigned
different games and hobbies than young boys are (Connor, Schackman, & Serbin, 1978; Sherman, 1978). Activities that aid and predict stronger visuospatial skills in children are made more available to boys rather than girls. Women tend to portray stronger overall performance in tasks reviewed by Serbin and colleagues, which they attributed to a greater social responsiveness and stronger sense of compliance. Attributes such as social
responsiveness have been described to be closely related to female gender expectations (Fagot, Leinbach, & Hagan, 1986). As a result, women to be more focused and willing to perform adequately, regardless of the task at hand. Research related to gender differences can therefore not isolate gender; the socialization and expectations associated with the
participant’s gender are of significant effect.
The lack of effect of gender on accuracy in this task is hard to attribute to one specific factor. As mentioned before, gender is a variable that is nearly impossible to isolate from social expectations – altering any effect it might have on performance. There is also the likely possibility that there simply is no significant difference in unconscious working memory accuracy. Conscious working memory tasks have shown significant, yet minimal differences between men and women (Duff & Hampson, 2001; Maeda & Yoon, 2013; Voyer et al., 2007, 2017), but it is uncertain whether or not these are transferable to unconscious working
data from tasks as the one evaluated in this thesis. Additionally, unconscious performance is a variable of which we have barely scratched the surface. Several mediators of whose existence we are perhaps not even aware yet could influence the interaction between gender and
unconscious performance. The aim of the experiment of which data has been analyzed, was to find within-subject differences. Due to this aim, no other demographics but age and gender have been collected. This limits any research into other factors that could influence
unconscious working memory performance in this task or the interaction between gender and performance. Demographic data collection could for example have included level of
education, as it is shown that working memory can affect and even predict educational achievements (Alloway & Alloway, 2010; Cowan, 2014).
Confounding factors on a more practical level should also be considered. Not all participants in the first experiment were right-handed or right-eye dominant. The Mondrians and unconscious stimuli were shown to opposite eyes to accommodate the participants that were left-eye dominant, which might have caused the results yielded in the first experiment to differ slightly from results yielded in other experiments. Any effect that different experiments could have on performance was however controlled for in the analysis by assigning
experiment the role of covariate. Another possible confound in the task performance is that it is unclear whether suppression of the target was successful. The contrast value of the target was altered based on the self-report of the participant. This subjective measure is sensitive to misunderstanding the question or instructions, causing the threshold for the stimulus to perhaps be too low or too high. It might therefore be that the participant respectively did not perceive anything – causing them to guess, leaving accuracy to chance – or perceived too much – causing them to consciously perceive.
A different limitation of the thesis, unrelated to the way data was collected, is that the mixed ANCOVA leads to an underpowered study. With two between-subject factors, i.e. gender and experiment, that respectively are divided in two and three categories, a set of 89 datapoints is simply not enough to generalize the study’s findings to the general population. Because the experiment did not significantly influence performance, the ANCOVA could be collapsed to a mixed ANOVA that only took gender and target side into account to increase power.
Reference list
Adam, J. J. (1999). Gender differences in choice reaction time: evidence for differential strategies. Ergonomics, 42(2), 327–335. https://doi.org/10.1080/001401399185685 Alloway, T. P., & Alloway, R. G. (2010). Investigating the predictive roles of working
memory and IQ in academic attainment. Journal of Experimental Child Psychology, 106(1), 20–29. https://doi.org/10.1016/j.jecp.2009.11.003
Andrejevic, M. (2012). Brain Whisperers: Cutting through the Clutter with Neuromarketing. Somatechnics, 2(2), 198–215. https://doi.org/10.3366/soma.2012.0057
Atkinson, R. C., Shiffrin, R. M., Atkinson, R. C., & Shiffrin, R. M. (1971). The control processes of short-term memory (Report No. 173). Social Sciences, 4(2).
Bona, S., Cattaneo, Z., Vecchi, T., Soto, D., & Silvanto, J. (2013). Metacognition of Visual Short-Term Memory: Dissociation between Objective and Subjective Components of VSTM. Frontiers in Psychology, 4. https://doi.org/10.3389/fpsyg.2013.00062
Bona, S., & Silvanto, J. (2014). Accuracy and Confidence of Visual Short-Term Memory Do Not Go Hand-In-Hand: Behavioral and Neural Dissociations. PLOS ONE, 9(3), e90808. https://doi.org/10.1371/journal.pone.0090808
Cashdan, E. (1998). Are men more competitive than women? British Journal of Social Psychology, 37(2), 213–229. https://doi.org/10.1111/j.2044-8309.1998.tb01166.x Cowan, N. (2014). Working Memory Underpins Cognitive Development, Learning, and
Education. Educational Psychology Review, 26(2), 197–223. https://doi.org/10.1007/s10648-013-9246-y
Der, G., & Deary, I. J. (2006). Age and sex differences in reaction time in adulthood: Results from the United Kingdom Health and Lifestyle Survey. Psychology and Aging, 21(1), 62–73. https://doi.org/10.1037/0882-7974.21.1.62
Duff, S. J., & Hampson, E. (2001). A Sex Difference on a Novel Spatial Working Memory Task in Humans. Brain and Cognition, 47(3), 470–493.
Goldstein, J. M., Jerram, M., Poldrack, R., Anagnoson, R., Breiter, H. C., Makris, N., … Seidman, L. J. (2005). Sex differences in prefrontal cortical brain activity during fMRI of auditory verbal working memory. Neuropsychology, 19(4), 509.
https://doi.org/10.1037/0894-4105.19.4.509
Jain, A., Bansal, R., Kumar, A., & Singh, K. (2015). A comparative study of visual and auditory reaction times on the basis of gender and physical activity levels of medical first year students. International Journal of Applied and Basic Medical Research, 5(2), 124–127. https://doi.org/10.4103/2229-516X.157168
James, W. (1890). The Principles of Psychology, Vol. 2. NY, US: Henry Holt and Company. Maeda, Y., & Yoon, S. Y. (2013). A Meta-Analysis on Gender Differences in Mental
Rotation Ability Measured by the Purdue Spatial Visualization Tests: Visualization of Rotations (PSVT:R). Educational Psychology Review, 25(1), 69–94.
https://doi.org/10.1007/s10648-012-9215-x
Magnussen, S. (2000). Low-level memory processes in vision. Trends in Neurosciences, 23(6), 247–251. https://doi.org/10.1016/S0166-2236(00)01569-1
Miller, G. A. (1956). The magical number seven, plus or minus two: some limits on our capacity for processing information. Psychological Review, 63(2), 81.
https://doi.org/10.1037/h0043158
Misra, N., Mahajan, K. K., & Maini, B. K. (1985). Comparative study of visual and auditory reaction time of hands and feet in males and females. Indian Journal of Physiology and Pharmacology, 29(4), 213–218.
Nyberg, L., & Eriksson, J. (2016). Working Memory: Maintenance, Updating, and the
Realization of Intentions. Cold Spring Harbor Perspectives in Biology, 8(2), a021816. https://doi.org/10.1101/cshperspect.a021816
Richet, C. (1886). LES ORIGINES & LES MODALITÉS DE LA MÉMOIRE: ESSAI DE PSYCHOLOGIE GÉNÉRALE. Revue Philosophique de La France et de l’Étranger, 21, 561–590. Retrieved from JSTOR.
Soto, D., & Silvanto, J. (2014). Reappraising the relationship between working memory and conscious awareness. Trends in Cognitive Sciences, 18(10), 520–525.
https://doi.org/10.1016/j.tics.2014.06.005
Suomala, J., Palokangas, L., Leminen, S., Westerlund, M., Heinonen, J., & Numminen, J. (2012). Neuromarketing: Understanding Customers’ Subconscious Responses to Marketing. Technology Innovation Management Review, 11.
Townsend, J. T., Ashby, F. G., & Ashby, F. G. (2014, January 14). Methods of Modeling Capacity in Simple Processing Systems. https://doi.org/10.4324/9781315802473-14 Voyer, D., Postma, A., Brake, B., & Imperato-McGinley, J. (2007). Gender differences in
object location memory: A meta-analysis. Psychonomic Bulletin & Review, 14(1), 23– 38. https://doi.org/10.3758/BF03194024
Voyer, D., Voyer, S. D., & Saint-Aubin, J. (2017). Sex differences in visual-spatial working memory: A meta-analysis. Psychonomic Bulletin & Review, 24(2), 307–334.
https://doi.org/10.3758/s13423-016-1085-7
Wong, A. L., Goldsmith, J., Forrence, A. D., Haith, A. M., & Krakauer, J. W. (2017). Reaction times can reflect habits rather than computations. ELife, 6.
Appendix A
Table A1
Breakdown of number of trials in Experiment 1 per category (conscious, unconscious or absent) and target side (left, right) (total = 222)
Trial category Target side left Target side right Total
Conscious 24 24 48
Unconscious 63 63 126
Absent - - 48
Total - - 222
Table A2
Breakdown of number of trials in Experiment 2 per category (conscious, unconscious or absent and target side (left, right) (total = 312)
Trial category Target side left Target side right Total
Conscious 32 32 64
Unconscious 112 112 224
Absent - - 24
Total - - 312
Figure A3. Experiment 3 Working Memory Task Trial
*Excluded from analysis
**In these trials, 2 out of 3 times the distractor was on the opposite side of the target (12 out of 18 or 24 out of 36). In 1 out of 3 cases, the distractor and target were on the same side.
Table A3
Breakdown of number of trials in Experiment 3 per category (conscious with distractor (consciousD), conscious with only the target (consciousT), unconscious with distractor (unconsciousD), unconscious with only the target (unconsciousT) or absent) and target side (left, right) (total = 222)
Trial category Target side left Target side right Total
