The contribution of general cognitive ability (g) to individual differences in reaction time (RT) on speeded cognitive tasks is well-established in psychology. Much less well-understood is how the components of central processing contribute to this correlation. Previous research suggests that rapid decision-making rather than perceptual speed accounts for the relationship by showing that perception can occur in parallel with other stages of processing but decision cannot. The current project applies Saul Sternberg’s method of additive factors to test the assumptions of this “serial bottleneck” in two different computer-based paradigms, both of which use stimulus presentation manipulated systematically in difficulty of perception (e.g., detection of a light-colored target against a light background) and in difficulty of decision (e.g., a comparison between similar values). Together with a short-form test of cognitive ability, this multi-leveled factorial design permits the detection of an interaction between either type of manipulation and cognitive ability, the existence or nonexistence of which would help to localize the origin of the g-RT relationship.
With 324 participants, we confirmed the predictive value of RT on cognitive ability (r = –.317, p < .001) as well as the main effects of both the perception (F1,323 = 49.76, p < .001) and decision (F3,323 = 79.41, p < .001) manipulations on response time. Additionally, we found that cognitive ability scores significantly mediate the relationship between speed of response and decision difficulty (F3,323 = 3.91, p = .009), but not perceptual difficulty (F1,323 = .012, p = .914) in a repeated-measures ANOVA, consistent with theoretical predictions made by the serial bottleneck that variance in g is localized primarily to the decision-making stage of information processing.
