Proactive interference in visual working memory, tested online and in the lab

The research question

Proactive interference is the tendency for information learned earlier to disrupt memory for newer information. Most studies of it use words. The authors asked whether the same interference affects visual working memory for which object was where, and whether resolving it takes measurable effort. In verbal tasks, pupil size, a real-time index of cognitive effort, has been shown to track interference (Johansson and colleagues, 2018).

The task

On each trial, four greyscale photographs of everyday objects appeared for 2 seconds at positions along an arc. A short question with two answer options followed (for example, "How many things are in a pair?"), which kept participants on task and limited rehearsal. One of the four objects then reappeared at the top of the screen, and participants pressed 1, 2, 3 or 4 to indicate where it had been.

Within each block of ten trials, the same two objects were shown on every trial, while the other two were new each time. When the tested object was one of the repeated ones, memories from earlier trials could compete with the current one (the interference condition). When it was a new object, there was nothing to compete with (the no-interference condition). The two trial types were mixed in random order and looked identical until the test, so participants could not study differently for one condition than for the other.

Running the experiment on Open Lab

Experiment 1 was built in lab.js and hosted on Open Lab. The 43 participants were recruited through Prolific, completed the study at home on their own computers, and responded by keypress. Each session had 3 practice trials and 100 test trials, with every trial lasting 14 seconds. Data from 34 participants met the authors' inclusion criteria, which excluded participants with repeated missed responses or below-chance accuracy. Pupil size was not measured online; the two follow-up experiments added it with an eye tracker in the lab.

What they found

In the online experiment, location memory was less accurate on interference trials than on no-interference trials (62.4% vs. 67.3% correct; t(33) = 4.33, p < .001, d_z = 0.74). The authors replicated this behavioural effect in the laboratory, both in Experiment 2A (36 participants analysed) and in the preregistered Experiment 2B (52 analysed). Pupil size was larger on interference trials in Experiment 2A but not in Experiment 2B. Across the three studies, the authors conclude that proactive interference reliably impairs memory for object locations, while the link between resolving it and effort, as indexed by the pupil, was not consistent.

In the researchers’ words