Abstract

Visual search is one of the most common paradigms used to study attention, for it allows the effective mimicking of tasks we perform naturally in our environment while maintain a larger amount of control over possible confounding variables. Although the paradigm in of itself has been quite beneficial to the field of attention research, the analyses that accompany it, focused predominantly on mean response times, and their positive slopes through increasing set sizes, have been demonstrated to be severely limited when describing the underlying architecture of search (parallel versus serial). In addition, the omnipresent skew of response times distributions nullify the possible interpretations typically associated with central tendency measures such as means.
We investigated how distributional analyses, which assess the entire response time distribution could accurately describe changes in response times through typical manipulations of visual search paradigms (set size, target presence and difficulty). We used the Weibull distribution, a left-bounded distribution to fit the response time data. Results demonstrated that search in of itself is not a dual architecture that changes under search difficulty, but a single mechanism that simply increases the duration of search when the difficulty conditions increased. The Weibull is therefore strongly recommended when analyzing response time data collected in visual search paradigms.