Authors:

BATTYE, G. & BARNARD, P.J.

Reference:

Year of publication:

2002

CBU number:

5160

Abstract:

Recent work has proposed that the role of the Anterior Cingulate Cortex (ACC) is to monitor conflict, and that feedback mechanisms mean that an increase in conflict in cognitive processing will lead to an increase in top down attentional control through excitatory activation of attended stimuli. We propose that a consideration of other paradigms, recently identified in fMRI studies to activate the ACC, provides evidence that conflict feeds back to trigger inhibitory attentional mechanisms rather than excitatory ones. We present a novel connectionist model of the attentional blink, based upon this theory and show that its behaviour is consistent with recent fMRI data. In recent work by Botvinick and colleagues, the occurrence of conflict has been identified as a common factor in studies that have shown ACC activation, and an explicit connectionist model of the ACC’s role in monitoring conflict and providing regulatory feedback to control processes has been proposed [1]. In two separate fMRI studies of the Eriksen flanker task and the Stroop task they have found ACC activity in accord with their model’s predictions of conflict monitoring activity in the Gratton effect and Stroop trial-type frequency effects respectively. In another area of attention research, the attentional blink (AB), current theories propose that an impairment in identifying a second target embedded in a rapid serial visually presented (RSVP) stream of distracters is triggered by conflict between the first target and the immediately following distracter. The role of the ACC as the location for conflict monitoring in the AB is supported by evidence from fMRI studies finding significantly increased ACC activation in two manipulations shown to produce an increased AB: increased temporal proximity of a target and the following distracter, and reduced target/distracter discriminability [2]. We have developed a connectionist model using the same general architecture and processing mechanisms as those used by Botvinick et al., demonstrating the occurrence of conflict in such RSVP paradigms. In contrast to preceding theories of how conflict might trigger a control signal that serves to increase excitatory signals to attended stimuli, our model relies on it to trigger inhibitory mechanisms so that attended stimuli may suppress their unattended competitors. We demonstrate that only the latter system is capable of replicating AB data, and that this system is also capable of capturing many of the empirical effects previously used as evidence in support of the former system. These studies inform the debate about the nature of attentional selection, and are especially pertinent in light of recent fMRI studies of the Stroop task [3], which provide empirical evidence for the role of inhibitory mechanisms in suppressing task-irrelevant information. The modelling work provides predictions about the functioning of conflict monitoring feedback which are testable in ongoing fMRI studies of the Eriksen flanker task. References 1. Botvinick M.M. et al. (submitted). 2. Marois, R. et al. (2000). Neuron, 28: 299-308. 3. Banich, M.T. et al. (2000). J. Cog Neurosci, 12, 988-1000.