Do Children with Autism Change Their Behaviour in Response to Volatility in the Environment?

Background:  

Adults can track reward probabilities across trials to estimate the volatility (or uncertainty) of the environment and use this information to modify their learning rate (Behrens et al., 2007). For example, in a stable environment, participants take account of outcomes over many trials, whereas in a volatile environment, they weight their recent experience more strongly than their distant experience. It has been suggested that individuals with autism may make less use of prior information (Pellicano & Burr, 2012), in which case, they may not change their behaviour in response to the statistics of the reward environment.

Objectives:

We investigated whether children with autism use information about the reward environment in a similar way as typically developing children to guide their decisions. Specifically, we hypothesised that children with autism would change their behaviour in response to volatility in the environment to a lesser extent than typically developing children.

Methods:  

We administered a developmentally appropriate version of Behrens et al.’s (2007) task to 35 children with autism aged between 6 and 14 years, and 36 age- and ability-matched typically developing children. Participants were shown a green and a blue pirate chest, each associated with a randomly determined reward value between 0 and 100 points, with a combined total of 100 points (Figure 1).  On each trial, the reward was given for only one stimulus. Participants initially completed a training phase where they passively monitored which of the two stimuli was rewarded on each of 20 trials before being required to estimate the ratio of rewards for the different colour chests.  In the test phase, participants were required to choose either the green or blue pirate chest using response pads, and were awarded points if they chose the correct stimulus (see Figure 1). Children completed this task under two conditions each consisting of 80 trials. In the stable environment condition, the ratio of the blue or green response being correct (and therefore yielding a reward) was fixed at 75:25. In the volatile environment condition, the ratio alternated between 80:20 and 20:80 every 20 trials. We estimated the learning rate for each participant by fitting a delta rule model, and compared these across conditions and groups.

Results:  

Overall, children increased their learning rate in the volatile condition compared to the stable condition.  Unexpectedly, however there was no effect of group and no interaction between group and condition.

Conclusions:

Children with autism use information about the volatility of the environment to guide their decisions to a similar extent as typically developing children. In decision-making, therefore, it appears that children with autism employ prior knowledge just as typically developing children do.