Memory representations contain a rich array of features, including perceptual and conceptual features of encoded stimuli. While prior research demonstrates that both feature types are independently represented in memory, available evidence suggests that recognition decisions (i.e., identifying one image as previously seen) depend primarily on conceptual features. However, memory judgments are a culmination of several processing stages: evidence accumulation, recognition decisions and subjective evaluations about memory quality. Whether perceptual and conceptual representations influence stages beyond recognition decisions, and whether these representations interact during these stages, remains untested. To address these questions, we conducted a series of experiments in which subjects performed forced-choice memory tests. In these tests, subjects encoded several images (e.g., the picture of a dog) and later had to discriminate between the encoded images and similar lures (e.g., the picture of a similar dog). Crucially, we employed computational models of vision to independently quantify the perceptual and conceptual similarity between the encoded images and their lures. Experiment 1 (N = 140) revealed that both perceptual and conceptual similarity impaired evidence accumulation, though only conceptual similarity decreased recognition accuracy. Experiments 2 (N = 254) and 3 (N = 315) revealed a similar effect, with conceptual similarity decreasing recognition accuracy. However, by directly testing interactions between both types of features, our results showed that when perceptual similarity between targets and lures was high, the detrimental effect of conceptual similarity on recognition accuracy was diminished -a reduction that was associated with participants increased reliance on more detailed-based retrieval strategies.
Morales-Torres et al. (Thu,) studied this question.