Abstract
How does the brain's spatial attention mechanism combine predictions about where an object (target) is likely to occur with predictions about the target's reward value? Some recent human findings suggest that these different forms of prediction bias attention independently (additive-bias hypothesis) whereas others, in keeping with economic notions of expected value, point to a multiplicative operation (expected-value hypothesis). To distinguish these conflicting theoretical alternatives, we used a simple incentivised visual search paradigm involving (1) viewing two placeholders for 300-400 ms that signalled the reward value for correctly identifying any letter target appearing in that location, (2) then viewing an additional central cue for 300 ms that signalled which placeholder was more likely to get a target, and then, (3) 100 ms after cue (but not placeholder) offset, identifying a target (H, N) versus a distractor letter (Z, K) that was presented within each placeholder for 100 ms. The validity of the cue varied in blocks. As expected, we found conventional spatial cueing effects (faster correct target identification for valid versus invalid spatial cues) that scaled with cue validity. Of interest was how reward value prediction derived from the continuously present placeholders modulated these cuing effects. Across two experiments, we found strong evidence that reward value predictions boosted cueing effects additively, even when it is suboptimal to do so. These findings refute theories that an expected-value computation is the singular mechanism underlying the deployment of endogenous spatial attention. Instead, it appears that spatial certainty and value associations independently bias selective visual attention.
Meeting abstract presented at VSS 2018