Periodic Reporting for period 1 - VISSATO (Visual Search and Cognitive Control of the Speed-Accuracy Trade-off)
Reporting period: 2018-09-01 to 2020-08-31
The visual environment contains many objects, yet only a few of them are behaviourally relevant. How we find relevant objects is a longstanding question in cognitive psychology. Research on selective visual attention has made tremendous progress in this area over the last forty years or so, underpinned by the visual search paradigm. In a visual search task, participants are asked to find a target among non-target objects on a computer screen. It is now well established that a set of basic visual features are initially extracted, prior to further scene analysis for relevant objects. Importantly, the accuracy of target-selection in visual experiments is subject to speed-accuracy trade-off (SAT), i.e. we are more likely to miss a target if we respond quickly. We exert a high degree of control over this trade-off, i.e. whether we favour speed over accuracy or vice versa, depending on our motivation. Such flexibility is believed to have been important for human survival, but nonetheless, little research has been conducted on how this control is exerted over visual search, much less its neural bases. With a focus on motion stimuli, we set out to develop a computational model of cognitive control for visual search, and to test the model's predictions for neural activity and behaviour under speed and accuracy emphasis.
We developed a distributed brain-circuit model of visual search, enabling us to simulate the performance of visual tasks under speed and accuracy emphasis. Under a single set of parameter values, the model accounts for neural signals and behavioural outcomes recorded during visual search tasks under speed and accuracy emphasis, offering a mechanistic explanation for the control for visual search, and making predictions for experimental testing. We mathematically reduced the model to a simpler form, enabling us to fit it to behavioural data from SAT tasks, accounting for individual differences in task performance. We further recorded electrophysiological and behavioural data from participants during visual search tasks with motion stimuli, characterising their behaviour and identifying neural signals of interest.
The main results of our research are that (1) the search for motion stimuli violates common characteristics of visual search more generally; (2) the control of SAT during visual search can be explained by the modulation of brain dynamics by two classes of signals, mediated by upper-layer and lower-layer cortical projections respectively; (3) these signals are well characterized by spontaneous, structured neural activity, resulting from unstable dynamics prior to stimulus onset on visual tasks; (4) the characteristics of these signals differ between individuals in ways that correspond to quantifiable strategies toward visual search under speed or accuracy emphasis; and (5) such strategies have behavioural manifestations that differentiate and explain participants' performance on related cognitive tasks.
The main results of our research are that (1) the search for motion stimuli violates common characteristics of visual search more generally; (2) the control of SAT during visual search can be explained by the modulation of brain dynamics by two classes of signals, mediated by upper-layer and lower-layer cortical projections respectively; (3) these signals are well characterized by spontaneous, structured neural activity, resulting from unstable dynamics prior to stimulus onset on visual tasks; (4) the characteristics of these signals differ between individuals in ways that correspond to quantifiable strategies toward visual search under speed or accuracy emphasis; and (5) such strategies have behavioural manifestations that differentiate and explain participants' performance on related cognitive tasks.
We have identified neural mechanisms by which control signals modulate visual circuitry under speed and accuracy emphasis, mechanisms by which these signals emerge prior to scene analysis, and mechanisms by which these signals guide selective attention in the visual field. This work provides a foundation for health research, perhaps most notably, the mechanistic bases of attention deficit disorder, cognitive decline in old age, and neuropathologies associated with cognitive impairment, such as schizophrenia. Our work can be exploited by clinical researchers through the testing of model predictions with patient groups in experiments on visual search and SAT. No website has been developed for the project.