Purchase this article with an account.
Justin Halberda, Tim Hunter, Paul Pietroski, Jeffrey Lidz; An interface between language and vision: Quantifier words and set-based processing. Journal of Vision 2008;8(6):234. doi: https://doi.org/10.1167/8.6.234.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
While limits of visual processing are interesting in their own right, these limits take on a deeper meaning where vision integrates with other cognitive systems. It is at this point that limits within vision become limits that can affect the whole of cognition. We present one such case. Subjects viewed briefly flashed arrays of 2–6 colors. Arrays always contained some number of blue dots among other dots. Subjects evaluated the verbal statement, “most of the dots are blue”. The concept MOST requires subjects to evaluate whether the number of blue dots is greater than the number of non-blue dots, but there are multiple ways to specify what counts as a ‘non-blue dot’: Hypothesis 1, these items are specified directly as the ‘yellow, green and red dots’; Hypothesis 2, these items are specified via a negation of the ‘blue dots’ (i.e. ‘non-blue dots’ are computed as ‘all dots’ minus ‘blue dots’). Hypothesis 2 is consistent with prevailing linguistic theory for the word ‘most’. We found that subjects behaved in accord with Hypothesis 2, selecting the blue dots and the superset of all dots. Psychophysical modelling revealed that subjects performed two operations: first taking the difference of two Gaussian numerical representations to compute the cardinality of the remainder set (superset − blue set = non-blue set) then comparing this computed Gaussian and the focused set to evaluate ‘most’ (blue set [[gt]] non-blue set). These two steps add error to the discrimination and the Weber Fraction for evaluating ‘most’ was twice as large as that for evaluating ‘more’ in a similar task. That is, which word subjects thought they were evaluating changed the observed Weber Fraction for this essentially visual discrimination. This difference highlights a case where non-visual cognition (lexical meanings) impacts vision and visual limits (tracking multiple sets) constrain later cognition.
This PDF is available to Subscribers Only