Found in translation: The role of response mappings for observing binding effects in localization tasks

ABSTRACT

According to action control theories, response and stimulus’ features are integrated into event files. Repeating any of an event file’s components retrieves the previously bound information, causing benefits for full repetition, but interference for partial repetition. Yet, such “binding effects” are absent in localization performance. By assuming sequential processing steps until response execution as assumed in visual search, we hypothesized that, for localization, participants can execute their response without the need to process target features. Hence, post-selective processing might be crucial for binding effects to emerge. Here, participants localized coloured targets appearing on one of four corners of a touchpad in two response conditions, namely, directly tapping on the target (direct response mapping), and tapping on the corner diagonal opposite to the target (translational response mapping). Only the translational response mapping yielded binding effects between localization response and colour. The direct response mapping instead showed an effect that is better explained by (non-spatial) Inhibition of Return or related change benefit effects. We conclude that an arbitrary response mapping – based on a translation of a spatial feature into a non-direct spatial response – can lead to binding effects even in localization tasks.

KEYWORDS:

• Action control
• attention
• perception
• stimulus-response binding
• response mapping

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data of both experiments is available at http://doi.org/10.23668/psycharchives.8271. Code for analysis of both experiments is available at http://doi.org/10.23668/psycharchives.8270.

Notes

1 Although the paradigm is different, the sequential design involving the systematic variation of repeating or non-repeating spatial and non-spatial features of a target stimulus is similar to the prime-probe sequences used in action control research (see Huffman et al., 2018; Schöpper, Hilchey, et al., 2020).

2 One study (Pratt & Castel, 2001) showed partial repetition costs in a localization task; this study was replicated by Huffman et al. (2018) with more power, with the results being in line with their overall interpretation.

3 Wolfe (1994, p. 232) discusses that detection responses can be based on the pre-attentive stage, potentially resulting from parallel search, but also mentions that such detection performance can be influenced by previous information (e.g., Bundesen, 1991; see also Found & Müller, 1996). Moreover, regarding location, detection performance in visual search tasks ("pop-out tasks"; e.g., Müller et al., 1995) produces a different outcome than detection performance in attentional orienting tasks (see, e.g., Hilchey, Pratt, & Lamy, 2019), namely, a benefit for a target repeating its location (e.g., Krummenacher et al., 2009) compared to changing its location (Huffman et al., 2018). We refer to such models, because they offer a serial processing architecture that can be transferred to the experiments at hand; however, we do not equate the (processes that lead to the) effects which we are investigating with effects that result from complex visual search displays.

4 Of note, in this design, location and response are fully confounded (c.f., localization tasks reviewed by Huffman et al., 2018; see also Hilchey, Antinucci, et al., 2019), so that location repetitions and changes equal response repetitions and changes. In turn, any observed binding pattern could be due to binding between location and colour, response and colour, or both. Of importance, binding between response and colour typically does not occur in simple localization performance (e.g., Schöpper & Frings, submitted). To avoid confusion, we will refer to the binding effect as being caused by binding between localization response and colour.

5 Error rates were close to ceiling; we therefore did not analyze the error rates in a manner comparable to reaction times. Overall error rates (i.e., the percentage of all incorrect probe responses given after all correct prime responses irrespective of condition) were higher in the translational response condition (1,00%) compared to the direct response condition (0,39%), t(38) = 4.38, p < .001, d = 0.70 (paired sample t-test, two-sided).

6 Location changes could be executed with the same (vertical changes) or the other thumb (diagonal and horizontal changes). Although event file retrieval can survive effector switches (Moeller et al., 2015), one might muse that using the same versus a different thumb could have had an impact on responding. We collapsed the horizontal and diagonal changes and conducted a 2 (response change type: same thumb vs. different thumb) x 2 (color relation: repeated vs. changed) ANOVA on probe reaction times of location changes separate for the direct and translated mapping. For the direct mapping, the main effect of change type was significant, F(1, 38) = 106.73, p < .001, ${\eta }_p^2$ = .74: Response changes were faster when executed with a different (492 ms) compared to the same thumb (535 ms). However, neither the main effect of colour relation, F(1, 38) = 2.65, p = .112, ${\eta }_p^2$ = .07, nor its modulation by response change type, F(1, 38) = 0.86, p = .361, ${\eta }_p^2$ = .02, were significant. For the translated mapping, a main effect of response change type emerged, F(1, 38) = 103.67, p < .001, ${\eta }_p^2$ = .73, also depicting a response change benefit when executed with a different (612 ms) compared to the same thumb (645 ms). Neither the main effect of colour relation, F(1, 38) < 0.01, p = .965, ${\eta }_p^2$ = .00, nor its modulation by response change type, F(1, 38) = 2.07, p = .158, ${\eta }_p^2$ = .05, were significant. To conclude, using the same thumb hindered overall responding, however, this did not have an impact on the effects of interest.

7 One could also expect IOR-processes in the translational response condition, as binding and IOR can work in parallel (Hommel, 2004), with binding effects potentially masking IOR (Hilchey, Rajsic, et al., 2018; Schöpper et al., 2022). However, we were interested in whether a translational response condition for localization performance yields a binding pattern and not whether such a response is (additively) affected by IOR-processes.

8 As with Experiment 1, error rates were close to ceiling (e.g., nine of 29 participants did not make errors at all after a correct prime). In total (i.e., error rate collapsed across conditions), participants gave 0.87% incorrect probe responses after a correct prime response.

9 Note that the color discrimination tasks in Schöpper, Hilchey, et al. (2020; Experiment 1: 412 ms, Experiment 2: 395 ms) were in fact even faster than the direct response condition in Experiment 1 of the current study (472 ms). However, we interpret this strong reaction time difference with some caution, as both studies used different response mappings and materials: Schöpper, Hilchey, et al. (2020) used two fingers mapped on two keys on a keyboard compared to our current use of two fingers mapped on four “keys” on a touch-monitor.