We agree with Taylor et al. (1) that there is value in examining the link between autism and social psychological skill (SPS) in terms of the subdimensions of autism. However, we note that even if autism has a multidimensional structure, the overall relationship we observed between autism and SPS remains valid (2). That is, the multidimensionality of a construct does not negate overall links between that construct and other variables, though such multidimensionality may help inform and provide additional insight into these links (3).

Although we agree that examining the subdimensions of autism can help shed additional light on our findings, in contrast to Taylor et al. (1), we feel that our data could not suitably test this question. First, as noted in our article, we were unable to examine whether the subdimensions of the assessed autism measure differentially predict SPS, because these subdimensions were not reliably observed (2, 4, 5). Second, unlike Taylor et al., we do not perceive a clear pattern regarding the individual autism items in terms of predicting SPS. Specifically, Taylor et al. claim that, “Broadly speaking, nonsocial traits [items on the autism measure] are linked to better performance, whereas social traits are negatively related to social psychological knowledge.” Although the data seem to trend this way, this pattern is not consistent across all of the items of the included autism measure. Of the 5 “social” autism items, only 2 items actually relate negatively to SPS, and these 2 items are almost identical: “When I’m reading a story, I find it difficult to work out the characters’ intentions” and “I find it difficult to work out people’s intentions.” Further, one of the other social items, “I know how to tell if someone listening to me is getting bored,” even positively predicted SPS (after being appropriately reverse-coded), although this link was quite small (see table 1 in ref. 1). Therefore, we are hesitant to conclude a social vs. nonsocial divergence, although, we acknowledge that these results are suggestive of Taylor et al.’s interpretation.

To be clear, we think it plausible that more-social aspects of autism negatively predict SPS, and less social, systemizing aspects of autism compensate for this decrease. Indeed, this aligns with our claim that heightened systemizing underlies the slightly positive link between autistic traits and SPS (3, 6). However, given the lack of a consistent pattern in terms of the nonsocial vs. social autism items, we think that examining whether the subdimensions of autism differently predict SPS may best be left to future research.

To summarize, first, the multidimensionality of autism does not negate the link we observed between autism and SPS. Second, examining the subdimensions of autism should provide greater insight into this link, yet we believe that our dataset is not well suited for this endeavor. Thus, we encourage future research to more thoroughly examine how the subdimensions of autism individually predict SPS. Doing so should complement our findings by showing exactly how autism is linked to SPS.