Dissociation in decision bias mechanism between probabilistic information and previous decision

• Main Authors: Yoshiyuki eKaneko, Katsuyuki eSakai
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2015-05-01
• Series: Frontiers in Human Neuroscience
• Subjects: Prefrontal Cortex; functional magnetic resonance imaging; Signal detection theory; Sensory detection; probability cue; decision history
• Online Access: http://journal.frontiersin.org/Journal/10.3389/fnhum.2015.00261/full

Target detection performance is known to be influenced by events in the previous trials. It has not been clear, however, whether this bias effect is due to the previous sensory stimulus, motor response, or decision. Also it remains open whether or not the previous trial effect emerges via the same mechanism as the effect of knowledge about the target probability. In the present study, we asked normal human subjects to make a decision about the presence or absence of a visual target. We presented a pre-cue indicating the target probability before the stimulus, and also a decision-response mapping cue after the stimulus so as to tease apart the effect of decision from that of motor response. We found that the target detection performance was significantly affected by the probability cue in the current trial and also by the decision in the previous trial. While the information about the target probability modulated the decision criteria, the previous decision modulated the sensitivity to target-relevant sensory signals (d-prime). Using functional magnetic resonance imaging, we also found that activation in the left intraparietal sulcus was decreased when the probability cue indicated a high probability of the target. By contrast, activation in the right inferior frontal gyrus was increased when the subjects made a target-present decision in the previous trial, but this change was observed specifically when the target was present in the current trial. Activation in these regions was associated with individual-difference in the decision computation parameters. We argue that the previous decision biases the target detection performance by modulating the processing of target-selective information, and this mechanism is distinct from modulation of decision criteria due to expectation of a target.