Abstract:
Eriksen’s flanker effect is a classic interference phenomenon that has been widely studied in cognitive and behavioural psychology. When a task-relevant stimulus (target) is presented with multiple task-irrelevant stimuli (flankers) surrounding that are incongruent with the target, people tend to make slower and less accurate responses to the task in contrast to the conditions when the target and the flankers are congruent with each other. Despite the extensive investigations, the precise neural mechanisms causing the flanker interference are still under debate. Here, I compared two major approaches that have been widely applied to model the flanker interference: the continuous-flow model (CFM) and the drift-diffusion model (DDM). Although these models have some superficial similarities, they posit different origins for the flanker interference. The CFM suggests that sensory information flows continuously from the sensory cortex into cortical areas responsible for response execution, so that interference can be conceived as competition between response channels. In contrast, the DDM emphasises interference arising at the sensory and cognitive levels. The present study tested these two models with neural measures to evaluate which model provides a more appropriate and neurologically plausible explanation to the flanker interference. For CFM, I recorded the lateralized readiness potential (LRP) as an indicator of relative response activation. In particular, I was interested in whether this would reveal evidence for early activation of the incorrect response when incompatible flankers were displayed (the “Gratton Dip”). For the DDM, I computed “drift rate” as an index of the accumulation rate of sensory evidence. To calculate the drift rate, I flickered the target and flanker stimuli at different frequencies and recorded the resulting steady-state visual evoked potentials (SSVEPs).
Results in the LRP failed to reveal the “Gratton dip” in response to incompatible flankers, but instead revealed a delayed peak in the incongruent condition. On the other hand, the drift rate calculated at occipital electrodes (centred on Oz) for this condition was notably lower than for the compatible condition. Both the delayed LRP peak and the slowed drift rate were significantly correlated with response time. Taken together, these data provide more support for drift-diffusion models in which flanker interference arises at sensory/perceptual levels of processing rather than at the level of response selection.