Attentional bias towards happy faces in the dot-probe paradigm: it depends on which task is used

ABSTRACT

Two recent articles [Gronchi et al., 2018. Automatic and controlled attentional orienting in the elderly: A dual-process view of the positivity effect. Acta Psychologica, 185, 229–234; Wirth & Wentura, 2020. It occurs after all: Attentional bias towards happy faces in the dot-probe task. Attention, Perception, & Psychophysics, 82(5), 2463–2481] report attentional biases for happy facial expressions in the dot-probe paradigm, albeit in different directions. While Wirth and Wentura report a bias towards happy expressions, Gronchi et al. found a reversed effect. A striking difference between the studies was the task performed by the participants. While in Wirth and Wentura, participants performed a discrimination task, they performed a location task in Gronchi et al. In Experiment 1, we directly compared the two versions of the dot-probe paradigm. With the discrimination task, the bias towards happy faces was replicated. However, the location task yielded a null effect. In Experiment 2, we found a cueing effect with an abrupt onset cue in both tasks. However, for the location task a congruence-sequence effect (a typical characteristic of response-priming processes) occurred. This result suggests that in the location task, attentional processes are confounded with response-priming processes. We recommend to generally use discrimination tasks.

KEYWORDS:

• Attentional bias
• dot-probe task
• spatial attention
• happy faces
• response priming

Acknowledgements

Development of the MacBrain Face Stimulus Set was overseen by Nim Tottenham and supported by the John D. and Catherine T. MacArthur Foundation Research Network on Early Experience and Brain Development. Please contact Nim Tottenham at [email protected] for more information concerning the stimulus set.

Disclosure statement

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

Notes

1 To be more precise, the authors found this surprising effect within a sample of younger participants, that is, a sample comparable to those of Wirth and Wentura (2020) and to the de facto standard samples in psychological experiments. This sample was compared with a sample of older participants (who showed a typical bias towards happy faces) in order to test a hypothesis relating to life-span psychology. (We will elaborate on this issue in the General Discussion.) Moreover, the authors varied the cue-target asynchrony (CTA) block-wise(100 ms vs. 500 ms).

2 Images from the following individuals were used: 01, 02, 03, 05, 07, 08, 09, 10, 20, 22, 23, 25, 32, 34, 36, 37.

3 Actually, we noted in the pre-registration (section on “secondary analyses”): “Participants’ individual trait-anxiety will be used as a covariate. However, we do not expect interactions between trait-anxiety scores and the attentional bias towards happy faces.” Indeed, using STAI as a (centred) covariate in the ANOVAs reported above yielded no significant effects involving STAI; all effects reported above (i.e., effects not involving STAI) were essentially the same. While writing this manuscript, it appeared to us that reporting simple correlations is more convenient for the reader.

4 Orgeta (2011) found a positive effect of M = 18 ms (see Table 2) with a cue-target asynchrony of 85 ms (17 ms cue presentation plus 68 mask), which was not tested independently. Even a conservative calculation of the standard error, however, indicates significance.

5 As an aside, since we only had to make minor adjustments to the programme code from Experiment 1 for Experiment 2, we can use Experiment 2 to demonstrate that the absolute null results of the location task in Experiment 1 were not due to erroneous code.

6 For the sake of transparency it should be noted that we first collected the data of the location group (because the location task was the target of our theoretical considerations). With some months delay, we recruited the discrimination task sample for control reasons (i.e., to show that no CSE occurs for this task). For the sake of convenience, we nevertheless treat task as an experimental factor in the following sections although recruitment did not follow a strict randomisation protocol. It was of course checked that samples did not overlap.

7 Ansorge and Heumann (2004) conducted onset-cue / onset-target experiments using the location task. In their Experiment 1b (which is most comparable to our location task control experiment), the overall validity effect had a size of d_Z = 2.31. Admittedly, the effect was clearly moderated by a further binary factor. Nevertheless, d_Z = 0.69 seems to be a conservative assumption.

Additional information

Funding

This research was supported by a grant from the German Research Foundation to Benedikt Wirth and Dirk Wentura (WI 5758/1-1; WE 2284/16-1).