Source-linked AI summary

Towards a Quantum-Like Cognitive Architecture for Decision-Making

Catarina Moreira, Lauren Fell, Shahram Dehdashti, Peter Bruza, Andreas Wichert

arXiv:1905.05176v2cs.AI

TL;DR

The paper addresses limitations of heuristic-based and computationally assumptive models of human decision-making. It proposes a quantum-like cognitive framework based on superposition, contextuality, and interference, and argues that the framework can represent more information than classical models while accommodating paradoxical findings and cognitive biases.

  • Problem

    Existing approaches rely on heuristic-based processing and strong assumptions about the human mind’s computational architecture, while classical models do not capture several cognitive biases and paradoxical decisions.

  • Method

    The paper proposes a unifying decision-making framework grounded in quantum mechanics, using quantum-like cognitive representations and interference without specifying heuristics.

  • Results

    The framework can represent more information than classical models and accommodate several paradoxical findings and cognitive biases reported in Lieder and Griffiths.

  • Takeaways & Limitations

    Quantum-like modeling offers a generalised approach to decision-making that supports alternative inferences for cognitive phenomena not captured by classical probability.

  • Takeaways & Limitations

    Associating quantum-like contextuality with the paper’s distribution is theoretically speculative, and contextuality’s consequences for cognitive probabilistic models remain little known and undiagnosed.

Abstract

from arXiv · show

We propose an alternative and unifying framework for decision-making that, by using quantum mechanics, provides more generalised cognitive and decision models with the ability to represent more information than classical models. This framework can accommodate and predict several cognitive biases reported in Lieder & Griffiths without heavy reliance on heuristics nor on assumptions of the computational resources of the mind.

1. Comment

The paper proposes a quantum-like cognitive architecture as an alternative to heuristic-based and computationally resource-assumptive decision models. It uses superposition, contextuality, and interference to represent perspective-dependent cognition, paradoxical decisions, and biases beyond classical models.

  • Motivation: Lieder and Griffiths model decision-making with bounded resource-rational heuristics and assumptions about computational costs and solution utility.Their approach targets paradoxical human decisions while relaxing the optimality criteria of Expected Utility Theory.
  • Framework: The proposed framework applies quantum-mechanical principles without specifying heuristics to accommodate the cognitive biases addressed by Lieder and Griffiths.It avoids assumptions about the computational architecture of the human mind.
  • Representation: Quantum cognitive models represent events as multidimensional vectors in complex Hilbert spaces, allowing judgments such as buying or not buying a car to remain in superposition.Individuals rotate basis states toward personal beliefs, producing different representations of the same decision scenario.
  • Quantum effects: Interference and contextuality make probabilistic inferences dependent on measurement conditions and decision-maker perspective.These properties support representations of conflicting, ambiguous, uncertain, and undecidable thoughts, while allowing probabilities that violate classical total-probability assumptions.
  • Contribution: The framework represents more information than classical models and can construct models that traditional classical approaches cannot capture.It is intended to model different minds with bounded cognitive resources and accommodate paradoxical decision scenarios and cognitive biases.
  • Conclusion: The paper concludes that quantum-like decision models can accommodate several paradoxical findings and cognitive biases while supporting perspective-dependent inferences beyond heuristics.The authors position the approach as a unifying framework based on quantum mechanics rather than a relaxation of normative theories alone.
Loading 1905.05176v2…