RESUMEN
There are a number of cognitive models of autism that aim to explain how mental processes are handled differently in the condition. These models make claims about the nature of cognitive function in people with autism, and suggest that these differences applied in social contexts lead to the characteristic behavioural patterns. However, it is difficult to study these cognitive differences directly because of the complexity of social situations. Studies of perceptual function are tempting as an alternative way to study cognition because it is far easier to control the conditions and the stimuli that participants are exposed to. This makes hypothesis generation and interpretation of results more objective and more convincing. However, the study of perception in autism hasn't been very productive in contributing towards a model of cognition in autism. In many areas there are studies reporting contradictory results, preventing arrival at a consensus about the largest unresolved issues in the area. These studies tend to be repeated multiple times, but continue to provide contradictory evidence that doesn't allow us to place confidence in any of the cognitive models. An approach to these issues is proposed, focusing on critical analysis of contradictory studies rather than the endless process of repetition. This allows previous studies to be interpreted more objectively and resolve conflicts, and guides the design of future studies in ways that avoid the pitfalls that have been identified. Both of these outcomes result in more productive work being done. The first example is in the study of motion perception in autism, where the use of non-identical stimuli has been problematic. On closer critical analysis, a fundamental aspect of the motion stimuli demonstrates that the contradictions might be expected based on the differences in stimuli used. Addressing this issue can move the field towards resolution. A second example is in the study of spatial frequency sensitivity. Here, poor study design has created results leading to an "eagle-eyed visual acuity" hypothesis of autism. Errors in the initial study are revealed, suggesting that the model should be abandoned. Finally, a general issue is the assumption of homogeneity of perceptual ability and genetics in autism, where the reality is that subgroups exist within the population of people with autism, and significant variation exists between them. The evidence for this is summarised and the issues that it creates explored.