For Control's Sake: An Experimental Study of One Model-Based Control Signature of Attention

Abstract

Internal models are crucial to the development of effective control systems, a tenet developed by a branch of engineering and mathematics called control theory. Internal models have been found in biological systems, such as the brain’s motor system, which constructs a ‘body schema’ of the body’s configuration for better control of movement. We apply control theory to attention, and propose that the brain constructs an internal model of the attentional process as well, called the ‘attention schema.’ Attention is a dynamic process that requires control, which would make it a likely candidate for utilizing an internal model. We test for this internal model in attention by looking for a signature of model-based control: savings when adapting to new environmental dynamics, or faster re-adaptation when the environmental dynamics are reintroduced after a washout period. Savings suggests the use of an internal model, because the faster adaptation with the re-introduction of environmental dynamics indicates that prior exposure to the dynamics, despite the washout period, benefits the controller. We suspect that the prior exposure to the dynamics had created an internal model of the state of attentional dynamics, so that when these particular dynamics reappeared, the model could simply be reinstated and attention appropriately biased, rather than the environmental dynamics being learned and accommodated from scratch. This signature is inspired by the model-based literature, and our study supports the hypothesis that attention may depend on model-based control. We used a cued-attention paradigm where participants learn to attend to a visual displacement between the location of the cue and the location of the target, where attention should be directed. Subjects were tested with 5 blocks of trials where each block was either a non-predictive block, meaning the chance of the target appearing on either side of the cue was .5 or a predictive block, where the target had a .85 chance of appearing on the valid side of the cue. These block types were interleaved, and the time course of adaptation in the second predictive block was compared to the time course of adaptation in the first predictive block. We looked for savings, or faster adaptation, in the second predictive block, and our analysis with 40 subjects provides preliminary evidence for this phenomenon. The observation of savings is strong evidence of an internal model of attention, which has not yet been proven, so the confirmation of findings in this study is critical for future work in attention.