Abstract
A widely believed consensus is that stronger task-irrelevant signals more severely disrupt task performance. If true, task-irrelevant signals that are too weak to be visible (subliminal) should less disrupt task performances than stronger supraliminal signals. To test whether this is the case, the subjects (n=16) were instructed to perform an RSVP (Rapid Serial Visual Presentation) task at the center of the display while ignoring a peripheral motion display. The degree of coherence in the motion display was varied from trial to trial (0, 5, 10, 20, 50, or 100 %). To our surprise, the RSVP task performance was the worst when the task-irrelevant motion coherence was as low as 5%. The results are at odds with the widely believed consensus. After the main experiment, we conducted another experiment to determine the coherent motion subliminal threshold. The same group of subjects as in the main experiment was instructed to report perceived coherent motion directions at various degrees of coherence (4AFC). The performance with the motion coherence at lower than 7.7% was around the chance level (25%) for all the subjects and therefore 7.7% is regarded as the motion coherence subliminal threshold. This result indicates that the 5% coherent motion, which most severely disrupted the RSVP task performance, was actually subliminal. One possible explanation for this apparently paradoxical result is that since the attentional system did not “notice” subliminal signals, the system was not triggered to filter the signals out. As a result, the subliminal signals might have more severely disrupted the task performance than supraliminal signals, which the attention system could “notice” and filter out.
Supported by NSF SBS-0418182, NIH R01 EY015980-01, Human Frontier Foundation RGP18/2004, and NSF CELEST.