A view to a click: Pupil size changes as input command in eyes-only human-computer interaction

In the field of Human-Computer Interaction (HCI), affective computation extends parameterization by including implicit user information, e.g. galvanic skin responses or heart rate variability. Physiological responses are generally involuntary and measurements primarily derive emotional features to differentiate cognitive and affective states. Biofeedback applications externalize covert physiological responses and allow interference by means of simple cognitive techniques. Here, signal responses do not necessarily indicate cognitive or affective dispositions but depict a user's intention. As demonstrated recently, even pupil diameter is subject to explicit control, albeit with individually varying degrees of success. The current study explores to what extent users can transfer implicit events (here: pupil size changes) into explicit input requests. In Experiment 1 we carried out biofeedback sessions to determine individual strategies to achieve cognitive control on pupil dilation. In Experiment 2 we applied explicit pupil dilations as input commands in a visual search-and-select task. Implementation of varying thresholds ensured the involvement of different levels of cognitive effort to exceed selection criteria. Results show that a short period of freely exploring the mechanisms of action enabled most subjects to achieve cognitive control on pupil dilation, even during increased tonic activation provoked by high task load. Hence, the associated dynamics no longer define automatic events but indicate the degree of explicitness to be imposed upon implicit processing. During pupil-based object selection, a strict criterion was associated with a high cognitive load, prolonged selection times and a notable number of time-outs, whereas a soft criterion correlated with fast selections and low cognitive demands. Together with the outcome of a user survey, the current findings suggest explicit pupil size changes constitute an appropriate selection mechanism for HCI even in the context of high cognitive workload.