The go/no-go priming task: automatic evaluation and categorisation beyond response interference

ABSTRACT

The response-window version of a go/no-go (GNG) response priming task is introduced using both evaluative (Experiment 1) and animacy decision (Experiment 2). In each trial a cue indicates which target category should lead to a key-press. The target is preceded by either a congruent or incongruent prime. The standard priming task was added as well. Both tasks yielded robust priming effects. However, they differed regarding a signature of response activation paradigms, that is, the Gratton effect (i.e. smaller priming effects following incongruent trials compared to congruent trials), which is present in the standard task but absent in the GNG task. This indicates that effects found with the GNG task are caused by different processes compared to the standard task. Experiment 3 tested an alternative account to explain priming effects in the GNG task. By manipulating response biases, Experiment 3 provides evidence for this account.

KEYWORDS:

• Evaluative priming
• go/no-go task
• response-window
• response interference
• Gratton

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Maria Clara P. de Paula Couto http://orcid.org/0000-0003-2595-9579

Dirk Wentura http://orcid.org/0000-0001-9907-498X

Notes

1. Our motivation at that time was that previous studies have shown that GNG procedures might have some advantages over binary choice tasks (Perea, Rosa, & Gomez, 2002). For example, a reduction of error variance that enters the standard evaluative priming task due to the need for binary categorisations and associated response selection processes (see, e.g. Borkenau & Mauer, 2007). This rationale does not play a role in this article.

2. Admittedly, a null result for animacy priming with the GNG-task would have constituted an even more interesting result, since it would have indicated uniqueness of the evaluative dimension. Of course, such a finding would have led to a different route of subsequent research since in this case the contrasting results of Experiments 1 and 2 would have had to be confirmed in a single experiment.

3. We will use differences in ER instead of RT differences. Typically, ER-based effects using a response-window technique should produce larger effects compared to RT-based effects in a non-response-window version of the paradigm (see, e.g. Draine & Greenwald, 1998).

4. Given the equivalence of an F-test with df_N= 1 with a t-test (here for the differences of the priming effects) and our specific prediction, a one-tailed interpretation is allowed (see, e.g. Maxwell & Delaney, 1990, p. 144).

5. To show this, we varied c while holding d′ constant, recalculated Hits and False alarms rates for each pair of d′ and c, and calculated – based on Hits and False alarms rates and the payoff matrix – the monetary outcome. On average, participants of the go bias sample received 6.14€; the optimal c on the ROC curve would have resulted in 6.15€.

Additional information

Funding

This research was supported by a CAPES/Humboldt Postdoctoral Fellowship from the Coordination for the Improvement of Higher Level Personnel (CAPES), and the Alexander von Humboldt Foundation (AvH) granted to Maria Clara P. de Paula Couto [grant number BRA/1153010].