Planning, prospective memory, and decision-making: three challenges for hierarchical predictive processing models

Basso, Demis · 2013 · Crossref

DOI: 10.3389/fpsyg.2012.00623

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 critiques Andrew Clark’s hierarchical predictive processing (HPP) model, arguing that while HPP effectively explains sensorimotor loops, it faces significant challenges when applied to higher-level cognitive processes: planning, prospective memory, and decision-making. The author contends that HPP is insufficient as a unified mechanism for these complex behaviors and requires either substantial expansion or collaboration with other non-hierarchical processing systems. Regarding planning, the paper challenges Clark’s definition of planning as deriving intermediate states from a future goal via Bayesian inference. The author argues this view is limited because initial plans are often vague sketches rather than precise sequences, particularly in underspecified tasks. Evidence from studies on the Traveling Salesperson Problem indicates that proficient planning relies on cognitive flexibility and the continuous refinement of strategies during execution, rather than the rigid summation of single actions predicted from a fixed goal. Furthermore, HPP struggles to account for the temporal dynamics of real-world planning, where actions are often delayed until specific conditions are met, unlike the immediate stimulus-response paradigms typically modeled by HPP. The commentary also highlights prospective memory (PM) as a critical challenge. PM involves executing a planned action only when specific contextual cues appear, requiring the inhibition of ongoing activities. The author notes that HPP models may incorrectly predict increased activation for prospective actions when exposed to lure stimuli that share partial similarities with activating cues, whereas human cognition successfully inhibits such false alarms. Additionally, HPP must explain how long-lasting intentions are maintained and inhibited over time, a capability not currently robust in models focused on immediate stimulus responses. Finally, the paper examines decision-making, citing Shadlen et al.’s accumulator model, which posits that decisions occur when accumulated evidence exceeds a threshold. While this model incorporates Bayesian elements, it avoids the strict reliance on posterior probability central to Clark’s framework. The author suggests that abstract decisions, which create rules rather than immediate actions, fall outside Clark’s current scope but are essential for a comprehensive theory of human cognition. In conclusion, the paper asserts that while predictive processing is evolutionarily efficient for basic input-output processes, it lacks the flexibility to manage the complexity of human higher-order cognition. The author proposes that future models must integrate large-scale, non-hierarchical mechanisms capable of recursively managing multiple processing systems to adequately explain planning, prospective memory, and decision-making.

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 2026-06-26; verification: verified.

Topics

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

• anticipation