A video and essay on neurosexism
I made a video on neurosexism, the flawed research being done regarding sex differences and why it's important. You can watch the video here [warning: I am much more articulate with the written word] but I had a few requests for the paper I wrote on this subject (that I use as reference for my video), so I am going to post that here as well! There are 'works cited' at the end of the paper, I recommend those papers (and Dawson's book) if you're interested in further reading on this topic.
The Relevance and Ethics of Exploring Sex-Difference
In this paper I will argue that since cognitive science is able to establish strong equivalence without implementation, neuro-scientific research on sex-difference uses flawed methodology in order to establish gendered inequivalence, ultimately begging the question between genuine difference and multiple realizability, while telling us very little about cognition generally, and much about the pervasiveness of modern sexism. First I will explain the part implementation plays in a strongly equivalent model of cognition. Second I will outline the ways in which sex-difference research tests vague (and often changing) hypotheses using flawed methods in order to find and support stereotypical sex-differentiated roles and behaviour. Third I will argue that these findings cannot tell us anything substantial about cognition in that they do not contribute to any model of strong equivalence, and cannot determine any genuine difference. Last I will outline what this line of research tells us about modern sexism.
The ultimate goal of cognitive science is to establish a strongly equivalent model of human cognition. And, surprisingly, neuroscience does not play an essential role in achieving that end. Two models of cognition are strongly equivalent if they are solving the same problem (the computational level), they are doing so using the same methods (the algorithmic level), and these methods are both based upon the same functional architecture (Dawson 172). So, as long as the functional architecture of the two models are equivalent, physical equivalence (the implementation level) is not required. The particular physical realization of some mental state is irrelevant to determining whether that mental state is functioning according to some theory of cognition. Moreover, the “same functional component can be realized in a wide variety of physical forms,” pointing to the current state of cognitive science in which the “functional theories of cognitive science and the biological theories of neuroscience are not mature enough to be synthesized” (ibid 174).
This is ignored by a line of neuro-scientific research which aims to find gendered inequivalence in the implementation level, and use these findings to make inferences about cognition generally. I will argue that this research is fundamentally flawed in four ways: (1) it uses vague and changing hypotheses and unreliable methods to find these differences, (2) the conclusions imply a genuine gendered difference in information-processing (rather than an instance of multiple realizability) despite no evidence to support that conclusion, (3) and, due to the current state of cognitive science, contribute very little to any strongly-equivalent theory of cognition. Finally, (4) this research perpetuates the gender binary, and the harmful stereotypes attached to it.
The line of research which examines sex-differences in the brain is performed through two methods of fMRI analysis: region-of-interest and within-group analyses. Robyn Bluhm examines these methods and finds that region-of-interest analyses, insofar as researchers are only looking at whether certain (pre-established) areas of the brain are active (according to some statistical threshold) in one group and not the other, are unable to explain how these differences relate to other parts of the brain, or how this activity is involved in any sort of behaviour. And, simply knowing that some area surpasses a statistical threshold of activity in one group and not the other, does not establish that it is more active in that group, or that the activity is relevant to any sort of information or emotion processing. A difference in activity in the two groups need not show that the region is more active in one group; in fact, this difference can often be attributed to other factors. For instance, if there is a greater variability of brain activity in the female group, it may result in the group’s average activity falling below the threshold. Furthermore, Bluhm finds that when using within-group methods, researchers often start out assuming that the same cognitive processes are occurring in both genders, only to conclude that because there is some kind of perceivable difference in activity, males must be using a different cognitive strategy. But as these analyses take place within-group and without reference to any direct comparison, their results are often misleading.
Bluhm examines a study which is indicative of this line of neuroscientific research: Shirao et al studied the brains of men and women within their own groups, and without comparison, and found a difference in activity - males surpassed some statistical threshold while women did not. This should not be immediately attributed to sex-difference as it can be explained by methodological flaws; for instance, a greater variability in women’s activity can easily affect whether a group reaches a statistical threshold set by the researchers. In fact, when the two groups were directly compared, there was no significant difference in activity. However, this perceived difference was not only used to infer a difference in activity, but also that entirely different cognitive processes were possibly being used. This common inference is problematic for two reasons: first, this analysis presupposes the same cognitive processes in order to compare activity, so concluding different strategies based on different activity is inconsistent with their own hypothesis. Second, and not surprisingly, these misleading results are a perpetuation of harmful stereotypes attached to sex differences; the researchers saw some perceivable difference and suggested that men were likely processing the task more cognitively while women were processing it emotionally. What is troubling is not just the research itself, but also the inferences that follow from it. In the case of the Shirao study, within-group analysis is problematic in that the results have been too readily accepted by researchers, and are used to come to conclusions which are inconsistent with the researchers own hypotheses.
Basically, neither method of research can tell us anything except that they have found some statistical difference in activity of some area of the brain between males and females. This research cannot tell us anything about how or why these differences might arise or exist; it cannot provide us with information as to how this activity relates to other parts of the brain. Cognitive science does not yet have a complete theory of general emotion processing without which any perceived differences are incomprehensible. These dubious methods of research reflect much about the uncritical way in which sex-difference is being explored in neuroscience but give us severely limited evidence for this alleged gendered inequivalence. And, given that physical details cannot yet contribute to any theory of strong equivalence, this research is evidence of the currently irrelevant role of implementation in cognitive science.
This research may be unable to tell us about how or why emotion-processing might differ between sexes, but can it conclude that there is a genuine gendered difference in information-processing? The researchers behind these studies certainly seem to think so. Bluhm found that despite the Shirao study’s results showing only that male brain activity in the left medial frontal gyrus passed a statistical threshold while female brain activity in the same region did not, the researchers used these results to claim that this could mean men are processing the task more cognitively than emotionally. This is a shoddy inference. Even if I grant the truth of their results, that there is a significant difference in activity in that part of the brain between the sexes, it does not follow that this difference is genuine; that is, not attributable to multiple realizability.
Multiple Realizability is the thesis that one mental state/property/event can be realized by many distinct physical states/properties/events. In terms of a computer, this thesis contends that “very few of the physical states of a computer are relevant to the kind of information processing that the machine is actually carrying out” (Dawson 27). Distinct brain states do not guarantee distinct mental processes as “only a small number of physical properties are germane to computing, and nature hasn’t been kind enough to tell us exactly what these properties are” (ibid). In this line of research, it is entirely possible that gendered differences in physical states do not point to “different cognitive strategies” as was suggested. In fact, it seems much more likely that this difference tells us something much less revolutionary: just that there is some difference in brain activity. These researchers are begging the question when they infer from some difference in activity that there is a difference in strategy. Especially considering that this inference is not supported by any evidence. The researchers provide no reason to believe that this particular region of the brain is importantly related to cognitive strategies; but, because it “lit up” more significantly in fMRIs of male brains than female brains, it is inferred that two entirely different strategies must be taking place.
This inference is even more contestable when it is considered within the context of cognitive science as a whole. As previously mentioned, the particular physical realization of some brain state cannot assist in providing any organizational principles regarding how the brain behaves as an information-processor. In other words, whether or not our left medial frontal gyrus is statistically significantly active when performing an emotion-processing task cannot tell us what theory of functional architecture is most plausible. It cannot help to grant strong equivalence between two models of cognition as only the algorithm and the representations on which it operates are relevant to its determination. Cognitive science has not yet found a set of physical properties that can be taken as instantiating any particular functional architecture. And, even if a single theory of functional architecture is accepted, the “physical complexity of the brain makes it extremely unlikely that all of our brains would implement this architecture in the same way” (ibid 175). So these neuro-scientific results cannot yet contribute to any model of strong equivalence, and they cannot tell anything about how we think, feel or behave. And research which concludes gendered inequivalence on the implementation level is currently irrelevant and misleading.
So, these dubious methods are commonly used to find sex differences and affirm sex stereotypes regardless of the current state of cognitive science in which any conclusive gendered inequivalence can tell us next to nothing about our information or emotion processing, behaviour, or any theory of strong equivalence. Despite contributing very little to any theory of cognition, this research does, in fact, tell us something important about neuroscience. The way that neuroscientists uncritically endorse a line of research which ultimately reinforces gendered stereotypes, such as the notion that women are biologically more emotional than men, tells us something about the pervasiveness of modern sexism. Modern sexism is largely based around the belief that equality has already been achieved; that the status quo no longer poses issues for women or that the social situation is supported by some kind of genetic encoding. In this context, modern sexism is perpetuated by research which denies the existence of a sexist bias, makes use of problematic methods, and is ultimately used to reinforce the belief that men and women differ in their emotional behaviour. Cordelia Fine argues that a “stronger weighting of genetic influence on behaviour is associated with greater moral tolerance of the social status quo” (290). When the status quo is institutionally stacked against one group, neuro-scientific ‘evidence’ which claims to find a biological basis for these stereotypical differences ought to be evaluated in terms of its relevance, accuracy, and ethical implications.
I hope to have outlined why this line of sex-difference research is, at least as of right now, irrelevant to the overall aim of cognitive science insofar as any perceived neurological sex differences cannot yet be distinguished as genuinely different (rather than an instance of multiple realizability). And even granting the perceived differences as genuine, this research cannot contribute to any general theory of emotion-processing. Cognitive scientists do not yet have an understanding of the true relation between mental processes and physical ones; without which, the relevancy of this line of research is unconvincing. As for the accuracy of their findings, Bluhm has shown why the methods used for this research are problematic, and the conclusions reached are likely a result of the modern sexism which fuels their biased hypotheses; revealing the possibility that “the differences were only observed because the scientists are so committed to their belief in sex/gender differences in emotion [processing] that they accept vague hypotheses and dubious methods in the first place” (Bluhm 11). Ethically, these findings are dangerous in that they reinforce sexist beliefs about the cognitive lives of men and women. These differences are used to explain why traditional gender roles are still prevalent; creating a biological foundation for stereotypes and denying the reality of oppression. The gender binary is problematic in itself; primarily because a rigid divide of abilities and social roles constructed according to sex cannot allow for any diversity of gender, which means that, legally and socially, the binary denies the existence of many non-conforming identities. In terms of those who do identify according to the binary, that is, men and women, this binary comes with a loaded system of roles and stereotypes which have, historically and presently, been used to justify the subjugation of women.
Neuroscientific research is potentially valuable insofar as implementational details will, ultimately, be a part of any fully developed cognitive science. However, our current understanding of cognition renders sex-difference research irrelevant. Without knowing how physical properties synthesize with an accepted theory of cognitive architecture, any differences found between male and female brains are irrelevant, misleading, and politically/socially precarious. In other words, the relationship between cognitive theory and actual neural circuitry will become clear only when strong equivalence is achieved; since implementation is not necessary toward that end, this line of neuro-scientific research does not currently contribute to cognitive science in any meaningful way. This means that the vague hypotheses, followed through with unreliable methodology, to reach conclusions which cannot be determined as genuine and perpetuate existing sex stereotypes ultimately tell us much more about the pervasiveness of modern sexism than anything to do with the goal of cognitive science.
Bluhm, Robyn. "New Research, Old Problems: Methodological and Ethical Issues in fMRI Research Examining Sex/Gender Differences in Emotion Processing." Neuroethics (2011).
Dawson, Michael R. W. Understanding Cognitive Science. Oxford: Blackwell Publishers Ltd, 1998.
Fine, Cordelia. "Explaining or Sustaining the Status Quo? The Potentially Self-Fulfilling Effects of 'Hardwired' Accounts of Sex Differences." Neuroethics 5 (2012): 285-294.