Discussion
People with autism spectrum conditions made fewer context-induced preference reversals than did neurotypical individuals. That is, they made more conventionally rational decisions. Our results accord with evidence of reduced loss/gain framing effects when people with ASC make choices between gambles (De Martino, Harrison, Knafo, Bird, & Dolan, 2008) and extend the extensive demonstrations of reduced sensitivity to global context in perceptual and cognitive tasks to a new domain: ASC participants were more likely than control participants to represent the value of each attribute or option in isolation, rather than being influenced by the other items in the choice set. This kind of reduced context sensitivity has traditionally been labeled weak central coherence—a diminished ability to integrate local information into a global gestalt (Frith, 1989). However, the original conception of weak central coherence does not capture enhanced choice consistency in a “high-level” decision task such as ours, in which there is no global percept. Rather, our data support more recent suggestions that autism is characterized by a wide-ranging enhancement of, or preference for, local information processing (e.g., Happé & Frith, 2006; Plaisted et al., 2003).
Why were people with ASC less susceptible than control participants to context effects in our choice task? There are many accounts proposing mechanisms for contextinduced preference reversals (see Howes, Warren, Farmer, El-Deredy, & Lewis, 2016, for a recent review). Two are of particular relevance to ASC. The first posits that choices are based on how readily they can be justified, “even when there is no overt need to justify to others” (Simonson, 1989, p. 159; see also Pettibone & Wedell, 2000). The target is better than the decoy on both dimensions (whereas the competitor is superior on only one), and this provides a reason to choose the target option, increasing its choice share (Simonson, 1989).
