Effects of stimulus features and instruction on response coding, selection, and inhibition: Evidence from repetition effects under task switching

Abstract

The coding of stimuli and responses is crucial for human behaviour. Here, we focused primarily on the response codes (or response categories). As a method, we applied a combined dual-task and task-switch paradigm with a fixed task-to-hand mapping. Usually, negative effects (i.e., costs) are observed for response category repetitions under task switching. However, in several previous studies it has been proposed that such repetition effects do not occur, if the stimulus categories (e.g., “odd” if digits have to be classified according to their parity feature) are unequivocally mapped to specific responses. Our aim was to test this hypothesis. In the present experiments, we were able to distinguish between three different types of possible response codes. The results show that the participants generally code their responses according to abstract response features (left/right, or index/middle finger). Moreover, the spatial codes were preferred over the finger-type codes even if the instructions stressed the latter. This preference, though, seemed to result from a stimulus–response feature overlap, so that the spatial response categories were primed by the respective stimulus features. If there was no such overlap, the instructions determined which type of response code was involved in response selection and inhibition.

Keywords:

• Repetition effects
• Response coding
• Task switching
• Attention

Acknowledgments

We thank Thomas Kleinsorge, Iring Koch, Nachshon Meiran, Mike Wendt, Marco Steinhauser, and Tobias Studer for their helpful comments on earlier versions of this manuscript. Furthermore, we are grateful to Cindy Eckart, Johannes Schult, and Karen Donath for collecting the data. This research was supported by a grant (Hu 432/9) from the Deutsche Forschungsgemeinschaft (DFG) to the second author.

Notes

¹ The participants in the study of Schuch and Koch (2004) actually responded verbally (“left”, “right”) to the first stimulus (S₁) and manually (left and right response buttons) to the second stimulus (S₂). Notice, however, that this neither questions the reasoning presented here, since it holds for R₂ (the manual response), which is relevant with respect to the repetition effects, nor precludes that the present logic can be applied to a situation with only manual responses for T₁ and T₂.

² As can be seen in Figure 2, the anatomical finger features are completely equivalent to the spatial inner/outer features of the response buttons. In the following, though, we always differentiate between the spatial (in the sense of left/right) and the finger-type categories.

³ Notice that the cue is completely redundant, since the task order is constant for a given participant throughout the whole experiment.

⁴ The same criterion was also used for analysing the individual data in Experiments 2, 3, and 4.

⁵ With respect to the task switch effects, we observed the expected switch costs for R₂. Whereas the same holds also for R₁, if the response times are considered, task switch benefits occurred in the error rates. Thus, with respect to the task switch effects, we observed a speed–accuracy trade-off in the R₁ results of the training blocks, which is not in line with our previous results (Hübner & Druey, 2006).

⁶ Note that in the training blocks of Experiment 2 we already used stimuli (traffic signs) for which it can be assumed that they do not have implicit spatial features. Thus, despite the invariance relations, which resulted (a) from performing task repetition blocks and (b) from a variable task-to-hand mapping in the task switch blocks, this may have been a further reason why the participants followed the instructions and used the finger-type categories in these blocks. Nevertheless, the invariance relations seem to be more crucial in this respect, since the participants also code the responses according to the finger features with spatial stimuli (digits), if the mappings are accordingly (i.e., invariant, see e.g., Hübner & Druey, in press; Watter & Logan, 2006).