whatever next? predictive brains, situated agents, and the...

EDITOR’ S NOTE A exceptionally large number of excellent commentary proposals inspired a special research topic for further discussion of this target article’s subject matter, edited by Axel Cleeremans and Shimon Edelman in Frontiers in Theoretical and Philo- sophical Psychology. This discussion has a preface by Cleeremans and Edelman and 25 commentaries and includes a sep- arate rejoinder from Andy Clark. See: http://www.frontiersin.org/Theoretical_and_Philosophical_Psychology/researchtopics/Forethought_as_an_evolutionary/1031 Whatever next? Predictive brains, situated agents, and the future of cognitive science Andy Clark School of Philosophy, Psychology, and Language Sciences, University of Edinburgh, EH8 9AD Scotland, United Kingdom [email protected] http://www.philosophy.ed.ac.uk/people/full-academic/andy-clark.html Abstract: Brains, it has recently been argued, are essentially prediction machines. They are bundles of cells that support perception and action by constantly attempting to match incoming sensory inputs with top-down expectations or predictions. This is achieved using a hierarchical generative model that aims to minimize prediction error within a bidirectional cascade of cortical processing. Such accounts offer a unifying model of perception and action, illuminate the functional role of attention, and may neatly capture the special contribution of cortical processing to adaptive success. This target article critically examines this “hierarchical prediction machine” approach, concluding that it offers the best clue yet to the shape of a uniﬁed science of mind and action. Sections 1 and 2 lay out the key elements and implications of the approach. Section 3 explores a variety of pitfalls and challenges, spanning the evidential, the methodological, and the more properly conceptual. The paper ends (sections 4 and 5) by asking how such approaches might impact our more general vision of mind, experience, and agency. Keywords: action; attention; Bayesian brain; expectation; generative model; hierarchy; perception; precision; predictive coding; prediction; prediction error; top-down processing 1. Introduction: Prediction machines 1.1. From Helmholtz to action-oriented predictive processing “The whole function of the brain is summed up in: error correction.” So wrote W. Ross Ashby, the British psychia- trist and cyberneticist, some half a century ago. 1 Compu- tational neuroscience has come a very long way since then. There is now increasing reason to believe that Ashby’s (admittedly somewhat vague) statement is correct, and that it captures something crucial about the way that spending metabolic money to build complex brains pays dividends in the search for adaptive success. In particular, one of the brain’s key tricks, it now seems, is to implement dumb processes that correct a certain kind of error: error in the multi-layered prediction of input. In mammalian brains, such errors look to be cor- rected within a cascade of cortical processing events in which higher-level systems attempt to predict the inputs to lower-level ones on the basis of their own emerging BEHAVIORAL AND BRAIN SCIENCES (2013), Page 1 of 73 doi:10.1017/S0140525X12000477 © Cambridge University Press 2013 0140-525X/13 $40.00 1

Upload: others

Post on 04-Jun-2020

2 views

Category:

Documents

0 download

Report

Download

Embed Size (px):

TRANSCRIPT

EDITOR’S NOTE

A exceptionally large number of excellent commentary proposals inspired a special research topic for further discussion ofthis target article’s subject matter, edited by Axel Cleeremans and Shimon Edelman in Frontiers in Theoretical and Philo-sophical Psychology. This discussion has a preface by Cleeremans and Edelman and 25 commentaries and includes a sep-arate rejoinder from Andy Clark. See:

http://www.frontiersin.org/Theoretical_and_Philosophical_Psychology/researchtopics/Forethought_as_an_evolutionary/1031

Whatever next? Predictive brains,situated agents, and the future ofcognitive science

Andy ClarkSchool of Philosophy, Psychology, and Language Sciences,University of Edinburgh, EH8 9AD Scotland, United Kingdom

[email protected]://www.philosophy.ed.ac.uk/people/full-academic/andy-clark.html

Abstract: Brains, it has recently been argued, are essentially prediction machines. They are bundles of cells that support perception andaction by constantly attempting to match incoming sensory inputs with top-down expectations or predictions. This is achieved using ahierarchical generative model that aims to minimize prediction error within a bidirectional cascade of cortical processing. Suchaccounts offer a unifying model of perception and action, illuminate the functional role of attention, and may neatly capture thespecial contribution of cortical processing to adaptive success. This target article critically examines this “hierarchical predictionmachine” approach, concluding that it offers the best clue yet to the shape of a unified science of mind and action. Sections 1 and 2lay out the key elements and implications of the approach. Section 3 explores a variety of pitfalls and challenges, spanning theevidential, the methodological, and the more properly conceptual. The paper ends (sections 4 and 5) by asking how such approachesmight impact our more general vision of mind, experience, and agency.

Keywords: action; attention; Bayesian brain; expectation; generative model; hierarchy; perception; precision; predictive coding;prediction; prediction error; top-down processing

1. Introduction: Prediction machines

1.1. From Helmholtz to action-oriented predictiveprocessing

“The whole function of the brain is summed up in: errorcorrection.” So wrote W. Ross Ashby, the British psychia-trist and cyberneticist, some half a century ago.1 Compu-tational neuroscience has come a very long way sincethen. There is now increasing reason to believe thatAshby’s (admittedly somewhat vague) statement is

correct, and that it captures something crucial about theway that spending metabolic money to build complexbrains pays dividends in the search for adaptive success.In particular, one of the brain’s key tricks, it now seems,is to implement dumb processes that correct a certainkind of error: error in the multi-layered prediction ofinput. In mammalian brains, such errors look to be cor-rected within a cascade of cortical processing events inwhich higher-level systems attempt to predict the inputsto lower-level ones on the basis of their own emerging

BEHAVIORAL AND BRAIN SCIENCES (2013), Page 1 of 73doi:10.1017/S0140525X12000477

© Cambridge University Press 2013 0140-525X/13 $40.00 1