Iconic versus working memory metacognition to evaluate the richness of perception: a registered report.
Nicolás Alejandro Comay, Guillermo Solovey, Pablo Barttfeld
Royal Society open science November 1, 2025 DOI: 10.1098/rsos.231805 (opens in new tab) via PubMed
Summary
AI-generated from the abstractIconic memory (IM) can hold more information than working memory (WM), but people are less aware of what they know in IM. In a registered experiment, participants performed a letter-identification task using either IM or WM. Replicating classic results, IM capacity exceeded WM capacity. However, when performance was matched, metacognitive sensitivity—the ability to judge the accuracy of one's own perceptions—was higher for WM. Contrary to the idea that people overestimate the richness of iconic memory, confidence was lower in IM than in WM. These findings suggest that IM supports identification with less consciously accessible information than WM, challenging the view that visual experience is rich and detailed.
Study at a glance
| Characteristics | Registered report Preregistered Peer reviewed |
|---|---|
| Keywords | Confidence Consciousness Iconic memory Metacognition Working memory |
| Key finding | Metacognitive sensitivity is higher for working memory than for iconic memory, even when performance is matched, and iconic memory is associated with lower confidence, not overconfidence. |
Abstract
The high capacity of human iconic memory (IM) has been taken as evidence that visual experience is rich and detailed, as introspection suggests. Opponents to this view argue instead that this impression is illusory, with conscious access being mostly limited to what we can attend to. To provide evidence of either view, in this registered report we compared metacognitive sensitivity levels between IM and working memory (WM) representations. The rationale was that, if pre-attentive IM information is as consciously accessible as attention-bounded WM information, metacognitive sensitivity should be comparable across the two memory systems. Replicating classic findings, our results showed that IM capacity exceeded WM capacity. Nevertheless, and despite matched performance, metacognitive sensitivity was higher in WM. We further examined whether reduced metacognition in IM could be explained by inflation-the tendency to overestimate perceptual richness-by comparing confidence levels across the two memory conditions. Pre-registered analyses showed no evidence of inflation, as IM was associated with lower confidence. Our findings suggest that IM supports identification with less consciously accessible information than WM, challenging rich-view interpretations of conscious perception.