Flanker and Simon effects interact at the response selection stage

Abstract

The present study aimed at investigating the processing stage underlying stimulus–stimulus (S–S) congruency effects by examining the relation of a particular type of congruency effect (i.e., the flanker effect) with a stimulus–response (S–R) spatial correspondence effect (i.e., the Simon effect). Experiment 1 used a unilateral flanker task in which the flanker also acted as a Simon-like accessory stimulus. Results showed a significant S–S Congruency × S–R Correspondence interaction: An advantage for flanker–response spatially corresponding trials was observed in target–flanker congruent conditions, whereas, in incongruent conditions, there was a noncorresponding trials' advantage. The analysis of the temporal trend of the correspondence effects ruled out a temporal-overlap account for the observed interaction. Moreover, results of Experiment 2, in which the flanker did not belong to the target set, demonstrated that this interaction cannot be attributed to perceptual grouping of the target–flanker pairs and referential coding of the target with respect to the flanker in the congruent and incongruent conditions, respectively. Taken together, these findings are consistent with a response selection account of congruency effects: Both the position and the task-related attribute of the flanker would activate the associated responses. In noncorresponding-congruent trials and corresponding-incongruent trials, this would cause a conflict at the response selection stage.

Keywords:

• Stimulus–stimulus congruency
• Stimulus–response spatial correspondence
• Simon effect
• Flanker effect
• Response selection

We wish to thank Alessandra Girardi for her help in data collection. Barbara Treccani and Carlo Umiltà were supported by grants from Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR) and the University of Padova.

Notes

¹ The colour Stroop task proper, wherein participants have to name the ink colour of names of colours, is classified as Type 8 ensemble in Kornblum et al.'s (1999) taxonomy. At variance with Type 4 ensembles, in this case the dimensional overlap also involves the response dimension, which refers to colours as well.

² A common finding with Simon tasks is that the magnitude of the Simon effect varies as a function of response speed (cf. Proctor & Vu, 2006). Two accounts have been put forward to explain this finding. According to De Jong et al. (1994), the changes in the magnitude of the Simon effect across RT bins reflect the time course of the corresponding response activation, with the smallest and largest Simon effects in the portions of RT distribution that correspond to the time points in which this response is least and most activated, respectively. According to an alternative hypothesis (Zhang & Kornblum, 1997), the distributional functions of the Simon effect reflect the statistical properties (i.e., a difference in variance) of the corresponding and noncorresponding RT distributions. Whatever the explanation, our distributional analysis does not support the temporal-overlap hypothesis proposed by Hommel (1993a)—that is, the idea that the activation of the corresponding response occurs very quickly and then decays or is inhibited. According to De Jong et al.'s time course explanation, we should conclude that the activation of the corresponding response increased, rather decreased, over time, whereas, according to Zhang and Kornblum's explanation, no variations over time of the corresponding response activation occurred, and our experimental manipulations affected the variances of the corresponding and noncorresponding RT distributions. More in general, our data do not support any hypothesis according to which the effect of congruency on the spatial correspondence effect is mediated by response speed—that is, the differences in the correspondence effect between congruent and incongruent conditions are due to differences in response speed between these two conditions. Indeed, such a hypothesis provides that (a) the two congruency conditions do yield two different response speeds, and (b) the RT distributional analysis shows a Simon effect function compatible with the observed differences in both response speed and Simon effect magnitude between the two congruency conditions (e.g., if faster RTs are observed in a certain congruency condition associated with a larger Simon effect, we should observe a decrease of the Simon effect as RT increases). That should be found regardless of whether this Simon effect function is the result of the time course of the corresponding response activation or of different variances of the RT distributions. Neither of these two predictions was confirmed by Experiment 1.

³ We thank Hagit Magen for having suggested this interpretation.

⁴ This hypothesis, as well as the temporal-overlap hypothesis proposed by Kornblum et al. (1999) and Hommel (1997a, 1997b), is based on the overlap in time between two events. However, these two hypotheses differ considerably. Kornblum et al.'s and Hommel's hypothesis refers to an overlap in time between the irrelevant spatial code and the response selection process. According to this hypothesis, the interaction between S–S congruency and S–R correspondence occurs because of a lack of temporal overlap between the spatial code and response selection in the incongruent condition, wherein a delay in the response selection stage is predicted. That should produce an underadditive interaction: The spatial correspondence effect in the incongruent condition should be either smaller than in the congruent condition (wherein the temporal overlap between the spatial code and response selection occurs) or even absent. The hypothesis here proposed to account for the discrepancy between the present and previous findings refers to an overlap in time between the irrelevant flanker spatial code and the irrelevant flanker colour code. This temporal overlap is not the cause but the prerequisite for the critical interaction to occur: The interaction is allowed to occur when there is a temporal overlap between the codes that correspond to the two irrelevant features of the flanker. Indeed, this interaction is caused by the interactive action of the two irrelevant codes at the response selection stage, and, in order to interact, the two coding processes need to have overlapping time courses. The distributional analyses of Experiment 1 RTs yield results that are inconsistent with the temporal-overlap explanation of the observed Congruency × Correspondence interaction: A time influence on the correspondence effect was found, but it was exactly opposite to what the temporal-overlap principle predicts. In contrast, these results are compatible with the hypothesis that the time window in which the two irrelevant codes are formed and wield their effects is crucial for the critical interaction to occur. The RT distributional analyses do not allow a direct test of this hypothesis: The time courses of the two irrelevant flanker codes should indeed be independent of RTs. Nevertheless, there is no inconsistency. When responses were very fast, very likely both irrelevant codes had not yet formed when the response was selected. Accordingly, neither the two compatibility effects nor their interaction were observed. In contrast, when responses were slower, an interaction between the two compatibility factors was found.