Purchase this article with an account.
Emily J. Ward, Gary Lupyan; Linguistic penetration of suppressed visual representations. Journal of Vision 2011;11(11):322. doi: https://doi.org/10.1167/11.11.322.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Linguistic labels (e.g. “chair”) appear to activate associated visual properties of the objects to which they refer (Lupyan 2008, Lupyan & Thompson-Schill, 2010). Labels can also inform visual tasks and offer performance advantages compared to nonverbal cues. Can hearing verbal labels make visible images that are otherwise invisible? In two experiments, we used anaglyph glasses and continuous flash suppression (CFS) to suppress images of objects from awareness. CFS involves presenting dynamic, high-contrast patterns to one eye and an object to the other eye and produces continuous suppression of the object. For each trial, participants heard either 1) a label corresponding to the suppressed object, 2) a label corresponding to a different object, or 3) white noise. Participants then viewed the pattern-object anaglyph and performed a simple detection task. If they detected any object, they were asked to verify its identity. We predicted that if labels activate visual information, hearing a label should “un-suppress” the object, but only if the label corresponds to the object.
Hearing a valid label prior to the object-detection task resulted in a significant increase in hit rate for simply detecting object presence, relative to baseline (an uninformative cue). Invalid labels resulted in a nonsignificant decrease relative to baseline. Signal-detection analysis showed a reliable increase in d′ following valid labels relative to baseline. We observed a similar pattern for verification responses. Analysis of correct-detection reaction times (RTs) revealed significantly shorter RTs following valid cues compared to invalid cues, and marginally longer RTs following invalid cues relative to baseline. A replication of the experiment using lower-contrast images to make detection more difficult yielded similar benefits of valid labels in both accuracy and RTs.
Labels may preactivate visual properties associated with the labeled object in a top-down fashion. This top-down linguistic assistance propels the image into awareness.
This PDF is available to Subscribers Only