Whatever after Next? Adaptive Predictions Based on Short- and Long-Term Memory in Visual Search

Conci, Markus; Zellin, Martina; Müller, Hermann J. · 2012 · Crossref

DOI: 10.3389/fpsyg.2012.00409

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This commentary by Conci, Zellin, and Müller addresses the temporal dynamics of predictive processing in visual search, offering a critique and extension of Andy Clark’s hierarchical predictive coding model. While Clark proposes that the brain functions as a prediction machine that minimizes prediction error to generate adaptive behavior, the authors argue that this framework lacks specificity regarding how predictions are adjusted over different time scales. The paper investigates whether the flexibility of predictive mechanisms varies depending on whether they are derived from short-term memory (recent experience) or long-term memory (statistical learning), suggesting that these distinct memory systems impose different constraints on adaptive behavior. The authors analyze existing empirical evidence from visual search tasks to differentiate between short-term and long-term predictive mechanisms. Short-term predictions are examined through the lens of intertrial priming, where recent target features and locations influence immediate spatial attention. Long-term predictions are evaluated using contextual cueing paradigms, which involve the implicit learning of stable associations between target locations and their surrounding spatial contexts. The commentary synthesizes findings from studies on statistical learning, bistable perception, and the persistence of attentional biases to characterize the adaptive properties of these two systems. The analysis reveals a fundamental asymmetry in the flexibility of these predictive models. Short-term predictions demonstrate high adaptability; for instance, observers can quickly adjust spatial attention based on recent history, though this may result in transient switch costs when targets change. In contrast, long-term predictions derived from statistical learning are characterized by rigidity and limited adaptive resources. Contextual associations are acquired rapidly but persist for long periods, showing slow extinction even when the target location changes. Furthermore, learning is typically limited to a single target location per context, and adaptation to changes occurs only if the change itself was initially predictable. This inflexibility suggests that long-term predictive models prioritize stability over immediate adjustment. The significance of these findings lies in refining the theoretical understanding of predictive processing. The authors conclude that predictive models should be differentiated based on their restrictions on adaptive processes. Short-term predictions require dynamic error minimization to account for frequent environmental changes, whereas long-term predictions operate under an assumption of environmental stability, making them less responsive to unpredictable shifts. This distinction implies that the degree of available error minimization varies across memory systems, resulting in a trade-off between flexibility and processing load. By integrating short- and long-term memory perspectives, the paper provides a more nuanced account of how the brain balances immediate adaptation with the efficiency of stable, learned expectations.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• anticipation