Select, then decide: Further evidence for separable selection and decision processes in short-term visual recognition

ABSTRACT

Recently, researchers have used evidence-accumulation models to analyse how performance in visual working memory (WM) tasks depends on both decision-making and non-decisional processes. One resulting claim is that selection of representations into the focus of attention and decision-making about those representations are separable processes, that this selection is facilitated by informative retro-cues, and that this process is evident in non-decision time when data are fit with a diffusion model. I attempted to address an alternative account of retro-cue effects on non-decision time: that they result from reduced uncertainty about non-memorial aspects of the task. In two 2AFC recognition experiments with retro-cues, participants had to choose between two probes – whose similarity to one another varied across conditions – to indicate which matched an item presented at a specified location in a sequentially-presented visual memory array. Crucially, the task minimized differences in uncertainty between the cue and no-cue conditions. Modelling showed a robust retro-cue effect in non-decision time, consistent with the advanced selection account of the retro-cue effect. Additionally, there were serial position and similarity effects, mostly evident on drift rate. These results support the contention that selection into the focus of attention is a separate process which facilitates stimulus use in decision-making.

KEYWORDS:

• Visual working memory
• retro-cue
• diffusion model

Acknowledgement

I thank Klaus Oberauer, Hsuan-Yu Lin, Hanna Fechner, and Marcel Niklaus for useful conversations pertaining to these experiments and the resulting data, Nathalie Rieser for her assistance with data collection, and Christian Olivers, Nathan Evans, and one anonymous reviewer for their helpful comments on an earlier version of the manuscript. Email: [email protected]

Disclosure statement

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

Data availability statement

All data and analysis scripts are available on the Open Science Framework at https://osf.io/zuskr/

Notes

1 Note that the range of possible values does not change, because in each case probes could have matched the cued memory stimulus, matched one of the uncued memory stimuli, or matched none of the memory stimuli. However, cues provided information about the relative probability of the probe matching the cued memory stimulus vs. matching one of the uncued memory stimuli, thus reducing overall uncertainty.

2 In HDDM, these variability parameters do not vary across conditions.

3 One reviewer suggested that the short ISI–250 ms – may also have played a part in the weak serial position effect. Though possible, note that Nosofsky and Donkin’s (2016) study used an ISI of only 150 ms, yet they found far stronger effects in the conditions with larger set sizes. This suggests that a short ISI is probably not the key factor here.

4 My thanks to an anonymous reviewer for this insight.