Abstract
Spatial information is often thought to hold a privileged role in visual working memory although it is unclear whether this always entails complete encoding of all available location information. In this study, we investigated to what extent spatial information about a visual memorandum is automatically encoded into VWM. Using a delayed estimation task, we directly assessed the quality of incidentally encoded location memory (experiment 1) and orientation memory (experiment 2). A surprise trial was introduced for either feature at a point when only the item’s color had been task relevant. This was followed by control trials to assess the memory quality when location or orientation had become task relevant along with color. We found the surprise trial performance to be significantly worse than the first control trial for both location and orientation, rejecting the notion that location encoding is completely automatic. Importantly, however, there was measurable location information on the surprise trial of experiment 1 while the surprise orientation report in experiment 2 was indifferentiable from a uniform distribution. Consistent with this difference, color memory precision was reduced after the surprise trial in both experiments, but less evidently so in the location-report experiment. We thus have convergent evidence that some location information was being encoded even when it was task irrelevant, but making the requirement of location report explicit induced the encoding of even more precise location information. All results were accompanied by data from pre-registered replication experiments. We reconciled our findings with previous studies suggesting automatic location encoding by emphasizing that automatic processes might be involved in encoding to various extents. Our study highlights the importance of acknowledging that the automaticity of memory encoding processes is not necessarily all-or-nothing and can exist along a graded scale.