The Influence of Prior Knowledge on Immediate Memory for Objects in ASD

Background: Recent evidence suggests that the perceptual experiences of individuals with ASD are modulated atypically by stored representations (priors), leading to a more accurate perception of incoming sensory information and less distortion by prior experience (e.g., Pellicano & Burr, 2012). It remains unclear, however, to what extent this difference in perception is a result of abnormalities in how experiences initially shape and update representations or in how representations are activated by sensory information. In the mainstream literature, Hemmer & Steyvers (2009) have developed an experimental paradigm that can shed some light on this issue. They showed that memory for the size of recently presented familiar stimuli (fruit, vegetables) is not only biased by the statistical properties of the stimuli presented during the experiment but also by the prior knowledge about the size of items that participants bring to the experiment.

Objectives: To shed further light on the interaction between prior knowledge and recent perceptual experience in ASD through the application of an adapted version of Hemmer & Steyvers’ (2009) paradigm.

Methods: 17 ASD and 19 age and IQ matched typically developing (TD) adults participated in two experiments. Experiment 1 probed participant’s long-term representations of the sizes of fruits by presenting photographs of these on a PC monitor and asking participants to resize them to what they considered to be the smallest possible, largest possible and average size of each item. Abnormalities in how prior knowledge is represented in long-term memory in ASD should lead to differences in the range and/or means of these sizes relative to controls. In experiment 2, 3 items were sequentially presented that could either be vegetables or control abstract shapes yoked in size to the vegetables. Participants were asked to remember the size of each item. Following each set, one item was re-presented in a new random size that participants needed to resize to the studied size. Importantly, the vegetable images included items that were either relatively small or large relative to their respective category mean (e.g., a small vs large radish). This design allows inferences to be drawn on how super-ordinate category and item-level representations of objects influence performance and comparison with the resizing of abstract shapes sheds light on how biases due to prior knowledge deviate from biases that develop during the course of the experiment.

Results: Experiment 1 revealed a tendency for ASD participants to report a greater range of sizes than TD participants, supporting the notion that prior knowledge may be represented atypically in ASD. In Experiment 2, participants with ASD were less accurate overall in re-constructing the sizes of to-be-remembered vegetables and shapes but both groups demonstrated similar levels of regression to the mean and biases that were induced by prior knowledge.

Conclusions: The results suggest that in ASD, the representations of recently viewed objects is slightly noisier or imprecise (Exp 2) and that this may lead to atypical representations of object properties in longer-term memory (Exp 1). The processes by which prior knowledge bias behaviour, however seem to operate typically.