Item-specific control of attention in the Stroop task: Contingency learning is not the whole story in the item-specific proportion-congruent effect

Abstract

A robust finding in the Stroop literature is that congruency effects (i.e., the color-naming latency difference between color words presented in incongruent vs. congruent colors) are larger for color words presented most often in their congruent color than for color words presented most often in incongruent colors. However, the cause of this item-specific proportion congruent (ISPC) effect is unclear, as it might be produced by either a conflict-adaptation strategy (e.g., prepare for conflict when the word RED appears) and/or a more general learning mechanism of stimulus-response contingencies (e.g., prepare to respond blue when the word RED appears). Thus far, attempts to directly dissociate the two processes appear to indicate no role for conflict adaptation, at least in situations in which contingency learning is also possible. We reexamined this conclusion in a Stroop task in which contingency learning and itemspecific conflict frequency were manipulated partially independently. In addition to a contingency-learning effect emerging for stimuli matched on conflict frequency, a conflict-adaptation effect also emerged for stimuli matched on contingency. The two effects also had different time courses, with the contingency-learning effect emerging early and remaining stable throughout the experiment and the conflict-adaptation effect arising later in the experiment. These results challenge not only the contingency-learning account of the ISPC effect, an account that denies the existence of a conflict-adaptation process, but also control accounts that assume that, although conflict-adaptation processes do exist, they are not used when contingency learning is also possible.

Working memory load dissociates contingency learning and item-specific proportion-congruent effects.

“…8Note that because the contingency-learning effect is relatively small in vocal responding (as shown in Experiment 1A) but a robust item-specific proportion-congruent effect is regularly observed when this response modality is used (as shown in Experiment 1B), one might conclude that, in vocal responding, the item-specific proportion-congruent effect might primarily reflect the action of a conflict-adaptation process rather than that of a contingency-learning process. However, the results of a recent item-specific proportion-congruent manipulation in our lab suggest a more cautious conclusion (Spinelli & Lupker, 2020a). In that experiment, the design permitted us to dissociate the independent contributions of contingency learning and adaptation to item-specific conflict frequency to the item-specific proportion-congruent effect.…”

“…As such, it is impossible to tell whether and how the contrast between color-specific and wordspecific information was resolved for those items. Thus, the comparison between mostly-congruent and mostly-incongruent incongruent words in Schmidt's and Hazeltine and Mordkoff's (2014) experiments may be one which is not diagnostic for adjudicating between conflict-adaptation and contingency-learning accounts of the item-specific proportion-congruent effect (see also Spinelli & Lupker, 2020a).…”

“…The fundamental difference between the control-based account and a contingency-learning account of the item-specific proportion-congruent effect is that the former assumes that participants in an item-specific proportion-congruent manipulation associate words with control processes (e.g., inhibit word reading upon presentation of the mostly-incongruent word GREEN) while the latter assumes that they associate words with specific responses (e.g., predict a yellow response upon presentation of the mostly-incongruent word GREEN). While both mechanisms might be used, researchers who have tried to directly dissociate the two accounts have mostly found support for contingency-learning processes (Hazeltine & Mordkoff, 2014; Schmidt, 2013a; but see Spinelli & Lupker, 2020a). For example, Schmidt (2013a) constructed a Stroop task in which item-specific conflict frequency and contingency learning were manipulated partially independently.…”

Bilingualism and executive attention: Evidence from studies of proactive and reactive control.

“…However, contingency learning is not always possible in MI lists, particularly when more than two colors are used and each word is presented in each of those colors (e.g., in an MI list, participants would not learn to associate the word RED with any response if RED appears in red, blue, green, and yellow equally often). Although there is increasing evidence that the role that contingency learning plays in both list-wide and item-specific PC manipulations is relatively minor (e.g., Spinelli et al, 2019, 2020; Spinelli & Lupker, 2020a, 2020b), it is not negligible either. Thus, we opted for a “symmetric” design, that is, one where contingency learning was possible in the MI block (e.g., participants could learn a RED-blue association) to the same extent that it was possible in the MC block (e.g., participants could learn a RED-red association).…”

“…Consistent with previous studies of the item-specific PCE, congruency effects were larger for MC items than MI items for both response latency and accuracy. Thus, participants selected appropriate item-specific control settings (i.e., focused attention for MI items vs. relaxed attention for MC items) without advanced preparation for these items (Bugg & Hutchison, 2013; Spinelli et al, 2020; Spinelli & Lupker, 2020a).…”