Controlled information processing, automaticity, and the burden of proof

Brian A. Anderson

Cognitive psychologists often distinguish between voluntary and involuntary/automatic processes in attention and cognitive control. Dedicated experimental paradigms have been developed to isolate involuntary information processing, but these paradigms tend to assume a rigid and inflexible process that is either stimulus-driven or built up through simple repetition. In contrast, voluntary information processing is often assumed when processing is in line with arbitrarily defined task-specific goals. Here I review evidence from multiple cases suggesting that ostensibly goal-directed cognitive processes may not be so voluntary and controlled. It is argued that automatic processes can be conditionalized to reflect the task relevance of the stimuli and selection history in a variety of ways, rapidly and flexibly adjusting in order to facilitate future goal-directed behavior. As a result, many studies assumed to have measured a voluntary cognitive process have likely measured an amalgam of voluntary and automatic processes, thus blurring the distinction between the two. Automaticity may be much broader and more sophisticated than has previously been thought, which has wide-reaching implications for our conception of human cognitive control.