Autism in the broader context of cognitive sex differences

Autism spectrum disorder (ASD) is characterized by difficulties with reciprocal social interactions, understanding and using (e.g., making eye contact) social cues, repetitive behaviors, and narrow interests (1). About 1% of individuals will meet current diagnostic criteria for ASD (2), with about a 4:1 ratio of males to females (3). Diagnostic criteria are categorical but the associated behaviors and cognitions fall along a continuum. Baron-Cohen et al. (4) proposed that ASD is at the extreme of the distributions of what they argued are two core human sex differences, and this is the focus of a recent study in PNAS by Greenberg et al. (5). These sex differences are called empathizing, which favors girls and women, and systemizing, which favors boys and men. The former represents sensitivity to the thoughts and feelings of others; the latter, a focus on identifying the rules that underlie how natural (e.g., weather) and mechanical (e.g., machines) systems work. From this perspective, ASD results from a combination of high systemizing and low empathizing, which is proposed to represent an extreme male brain (EMB). The EMB hypothesis integrates sex differences in brain and cognition with the core features of ASD and provides a parsimonious explanation of the overrepresentation of males with this disorder. The sex differences in empathizing and systemizing, in turn, are hypothesized to represent ancestral sex differences in the division of labor, specifically the greater parental care of women, and hunting and tool making in men (6).

The study of sex differences in brain and cognition has a long and fraught history, but a consensus has emerged for many cognitive sex differences (discussed below), and brain imaging studies are converging on the functional and structural brain systems underlying the cognitive differences (7, 8). Debates now center on the magnitude of the sex differences and the mix of biological and experiential factors that contribute to them. Evolutionary models of sex differences are particularly contentious (9) but can be constructed based on human cognitive universals and cross-cultural studies of sex differences. A focus on universals helps to distinguish sex differences that arise independent of context—even if context influences their development and expression—from those that are highly dependent on formal schooling, such as reading and mathematics.

One implication of the EMB hypothesis is that ASD and associated sex differences emerge from these cognitive universals. Thus, considering what these universals might be, as well as any associated sex differences in them, provides a broader context for thinking about ASD generally and the EMB hypothesis specifically. One such framework is shown in Fig. 1 (10), whereby human cognition is organized around folk psychology (competencies and cognitive representations concerning the self, other people, and groups of people), folk biology (knowledge of other species as used in hunting and foraging), and folk physics (competencies associated with navigation and tool use). These represent the basic cognitive architecture and biases that allow people in traditional contexts to develop and maintain relationships, live in social groups, extract resources from the ecology, and navigate in and manipulate the physical environment (e.g., build shelters). These competencies are not the same as the knowledge base in the academic fields of psychology, biology, or physics, and thus there is no direct relation between sex differences in folk domains and any differences in these academic fields.

Fig. 1.

Human universal cognitive abilities are organized around folk psychology (organization and processing of social information), folk biology (organization and processing of information about other species), and folk physics (systems for navigation and knowledge about nonliving things, including use of tools). Schema refers to specific knowledge in the domain. Girls and women generally excel in the processing of individual-level social information (green) that supports the formation and maintenance of dyadic relationships, whereas boys and men excel in movement in (navigating) and representation of large-scale space and in other aspects of folk physics (blue) (10).

In any case, girls and women have advantages in the areas of folk psychology that support the formation and maintenance of dyadic relationships (green highlight in Fig. 1). These include advantages in language fluency and comprehension, interpreting facial expressions and gestures, and correctly inferring the thoughts and feelings of other people, known as theory of mind (e.g., refs. 11 and 12). Girls and women have small to moderate advantages in each of these individual areas, but have about 1 SD advantage when cues are combined, as would happen in a natural social interaction. These sex difference were the focus of Baron-Cohen et al.’s (4) empathizing measures, although these measures produce sex differences that are less than one-half of that found for combined social cues and, thus, underestimate the differences. Boys and men typically have advantages in folk physics (blue highlight in Fig. 1), including various aspects of spatial cognition and tool use (e.g., ref. 13). These sex differences were the focus of Baron-Cohen et al.’s (4) systemizing measures, although these measures also seem to capture aspects of occupational interests in modern societies (14). Again, the sex difference in systemizing measures underestimates the sex differences in some areas of folk physics.

In keeping with a universal and potentially evolved basis, anthropologists are beginning to find many of the same sex differences in people living in traditional societies. For example, Cashdan and colleagues (15,16) demonstrated that the male advantage in spatial cognition found in Western populations is also found in traditional cultures, including Mayan farmers (Yucatan Peninsula), Hadza hunter-gatherers (Tanzania), and Twe and Tjimba horticulturalists and gatherers (Namibia). Among the latter, men with better spatial abilities and larger navigational ranges fathered more children with more partners than did their peers, in keeping with positive evolutionary selection for these abilities in men. Still, these types of studies are in their infancy and much remains to be determined. The key point is that there are reliably documented sex differences for many brain, cognitive, and behavioral traits, and these will likely prove to be useful in the study of various neurological and other disorders (17), including autism, as proposed by Baron-Cohen et al. (4).

In PNAS, Greenberg et al. (5) provide the largest evaluation to date of the EMB hypothesis. Two samples totaling 685,000 people were administered empathizing and systemizing measures along with a screener for ASD. Consistent with prior studies, there were moderate sex differences on the empathizing and systemizing measures, and roughly 5% of the self-selected sample who reported an ASD diagnosis scored lower on the empathizing and higher on the systemizing measures than did neurotypical individuals. One indicator of EMB is a large difference (about 2 SDs) between one’s relative (to the mean) empathizing and systemizing scores. About 11% of the men and 8% of the women with ASD met this criterion, compared with 4% and 2% of the neurotypical men and women, respectively. These results and others are consistent with the core predictions of ASD as a manifestation of EMB.

At the same time, there are many issues that remain to be resolved regarding how to best conceptualize sex differences in brain and cognition and integrate these with studies of ASD. As noted, the empathizing and systemizing measures underestimate sex differences in some areas of folk psychology and folk physics; thus, Greenberg et al.’s (5) findings do not fully incorporate cognitive sex differences into our understanding of ASD. Moreover, the core dimensions of ASD—poor social interactions and poor understanding of social cues along with repetitive behaviors and narrow interests—do not appear to be part of a single underlying deficit. Rather, individuals with ASD are heterogeneous and, at behavioral, cognitive, and genetic levels, each of these core deficits is only modestly correlated (18), meaning that it will be difficult to capture the fundamental deficits of ASD with a single index such as the difference between empathizing and systemizing scores (5), or with a single concept such as EMB (4). Even under the umbrella of systemizing, high-functioning individuals with ASD show uneven competencies. They are often average or better on some forms of spatial cognition, such as visual tracking, as predicted by the EMB hypothesis, but often show deficits on other forms of spatial cognition, such as difficulties forming allocentric (i.e., “birds eye”) representations of large-scale space and in systematically exploring this space (19, 20).

Despite these caveats, Baron-Cohen et al. (4) and Greenberg et al. (5) are correct in arguing that consideration of sex differences that emerge in neurotypical populations will enhance our understanding of neurological disorders that show a sex bias (17), including ASD. The findings of Greenberg et al. (5) confirm that men and women with ASD show a male-typical pattern of cognitive abilities and interests. Future studies of ASD using measures that more fully capture cognitive sex differences will likely provide a more complete understanding of the strengths and weaknesses of these individuals and a better understanding of the heterogeneity within ASD populations.