Project Awarded: $21,000
On the predictive coding (PC) view the mind engages in Bayesian hypothesis generation, testing and revision with the aim of minimizing the mismatch between its predictions and the sensory evidence (prediction error). Though the bulk of the empirical support for the theory lies in the perceptual domain (e.g. Hohwy, et al. 2008; Huang & Rao, 2011; Stefanics et al. 2014), the
Bayesian framework also promises to deliver a comprehensive theory of attention that falls out of the tools employed in the perceptual theory, without the need for positing additional machinery.
According to this proposal, in all instances of attentional behavior the selectivity of attention is explained on the basis of expected precision; attention selects a signal for further processing because the brain expects that signal to deliver a more precise or consistent set of data. The PC theory of attention is thus committed to the claim that high precision expectations are driving attention in all its instances.
However, there has been negligible empirical investigation of this claim to date (Egner, Monti & Summerfield 2010; Jiang, Summerfield & Egner 2013; Kok et al. 2011). We propose to address this gap by using affect-biased attention (ABA) as a test case for the theory, because it is a well studied phenomenon that at least prima facie poses a challenge to PC’s claim. ABA is attention to stimuli that are affectively salient, i.e. stimuli that stand out because the agent associates them with reward or punishment (Todd, Cunningham, Anderson, & Thompson, 2012). ABA has properties of both bottom-up and top-down attention biasing, and so is not easily captured by the PC model. Affectively salient objects can capture attention even when they are not physically salient (Niu et al. 2012a, 2012b; Awh et al. 2012), preventing straightforward assimilation to predictive coding’s treatment of bottom up attention. In addition, affectively salient objects can capture attention when they are not task relevant (Awh et al. 2012; Todd et al. 2012), preventing straightforward assimilation to PC’s treatment of top down attention. To assess the explanatory adequacy of the PC theory of attention, therefore, we propose to tease apart precision expectation and affective salience experimentally, probing their respective effects on the behavioral responses evoked by visual stimuli. By manipulating these two variables independently, it should become clear whether or not PC can accommodate affective biases on attention.