Abstract
A central proposal of embodied cognition is that perceptual features constitute an essential part of the conceptual representation. Indeed, experimental studies of everyday metaphors such as ‘happy is up’ and ‘sad is down’ have revealed how processing concepts can generate spatial biases toward locations compatible with the concepts’ meaning. An important question that has remained unanswered is whether spatial association between positive and negative valence (e.g., happy/sad) is in fact conceptual in nature. That is, the observed visuospatial biases may have been simply due to perceived stimulus valence and, as such, non-conceptual. To disentangle conceptual and perceptual processing, we tested the ability of conceptual (affect words) and perceptual (affective faces) stimulus valence in causing visuospatial bias above/below fixation. In each trial, observers were presented with a valenced (or neutral control) stimulus, which was categorized either based on valence (positive vs. negative) or a perceptual feature (upright vs. upside down), followed by a visual target above or below fixation, and made a speeded keypress response to the peripheral target. First, whereas spatial biases were always observed following the conceptual stimuli, with perceptual stimuli, spatial biases were observed only when subjects categorized the faces based on valence. Second, inverting the perceptual (face) stimuli, which induced conceptual ambiguity along the vertical axis (e.g., above fixation is down to an inverted face), reversed the direction of spatial bias, again consistent with the conceptual nature of the effect. Finally, we varied gaze direction of the same face stimuli and found no interaction between the valence-induced bias and the effect of upward/downward gaze cues, suggesting that the mechanism underlying the valence-based spatial bias differs from those responsible for covert shifts of spatial attention. Thus, the perception of positive and negative valence alone cannot generate spatial bias and that conceptual processing is, indeed, necessary for visuospatial biases.
Meeting abstract presented at VSS 2012