Aaron Cochrane¹, Takeo Watanabe¹, Yuka Sasaki¹; ¹Brown University

Human vision is shaped by our environments through both explicit goal-oriented and implicit goal-irrelevant interactions with stimuli, each of which can induce plasticity. The mechanisms underlying implicit perceptual learning have far-reaching implications due to the presumed ubiquity of such learning in development, rehabilitation, and occupational contexts. However, these mechanisms remain poorly understood. Here, we investigated the possibility of volitional influences on implicit learning. Specifically, we utilized a recently-developed biofeedback method for modulating pupil size, associated with noradrenergic activity, to investigate how such modulation affects implicit perceptual learning. Participants (n = 8) were instructed on some trials to dilate and on others to constrict their pupils, with real-time visualization of pupil diameter displayed in front of a random dot-motion stimulus. A third trial type (control) involved passive viewing of the stimulus without pupil-size manipulation. Over seven days of training, participants improved their ability to both dilate and constrict their pupils. Unbeknownst to participants, dilation, constriction and control trials were each associated with different directions of perithreshold coherent motion, separated by 120 degrees. No participants reported noticing coherent motion during the biofeedback procedure. From pre-test to post-test on a 3AFC motion discrimination task, participants improved their perceptual threshold for motion direction associated with pupillary constriction as well as for the control direction. In contrast, thresholds for the direction associated with pupillary dilation did not improve, and the change in threshold associated with dilation was reliably smaller than for either constriction or control. These results are compatible with previous findings that increased noradrenergic activity associated with pupil dilation may enhance goal-relevant stimulus processing in a way that suppresses goal-irrelevant features. Such volitional control trained via biofeedback could be useful in contexts where implicit, goal-irrelevant learning is undesirable, such as occupational training or sequential tasks wherein retrograde interference may be detrimental.

Acknowledgements: We thank funding to YS and TW: NIH (R01EY019466, R01EY027841, R01EY031705, P30GM149405); NSF-BSF (BCS2241417)