Criteria for an effective theory of consciousness and some preliminary attempts

https://doi.org/10.1016/j.concog.2003.09.002Get rights and content

Abstract

In the physical sciences a rigorous theory is a hierarchy of descriptions in which causal relationships between many general types of entity at a phenomenological level can be derived from causal relationships between smaller numbers of simpler entities at more detailed levels. The hierarchy of descriptions resembles the modular hierarchy created in electronic systems in order to be able to modify a complex functionality without excessive side effects. Such a hierarchy would make it possible to establish a rigorous scientific theory of consciousness. The causal relationships implicit in definitions of access consciousness and phenomenal consciousness are made explicit, and the corresponding causal relationships at the more detailed levels of perception, memory, and skill learning described. Extension of these causal relationships to physiological and neural levels is discussed. The general capability of a range of current consciousness models to support a modular hierarchy which could generate these causal relationships is reviewed, and the specific capabilities of two models with good general capabilities are compared in some detail.

Introduction

Investigations of consciousness have ranged from attempts to develop computational theories to the more limited goal of identifying the neural correlates of consciousness. One major element has been an argument around whether a scientific account of consciousness is possible, or whether there could be an explanatory gap in understanding how physiological processes generate the “what it is like” of consciousness.
The paradigm for a successful scientific theory is the physical sciences. This paper reviews some of the characteristics of physical theories as a model for what is needed for consciousness. Such theories establish hierarchies of description on many levels of detail in which causal relationships on one level can be mapped into causal relationships on any other level. It is then argued that the requirement to learn without undesirable side effects on prior learning will have forced the brain into the form of a modular functional hierarchy which has properties making it an effective vehicle for a hierarchical scientific theory.
The causal relationships implicit in generally used definitions of access and phenomenal consciousness are then made explicit. The equivalent but more detailed causal relationships observed in explicit and implicit mental processes are reviewed. A range of architectural models of consciousness are evaluated for their potential capability to support the causal relationships at the highest level, and two models are found to have such potential. The ability of these two models to support the causal relationships found in explicit and implicit mental processes are then considered in more detail, and the consistency of the models with psychology and physiology briefly evaluated.
The conclusion is reached that models of this type have some potential to become scientific theories of consciousness with explanatory capability analogous with theories in the physical sciences.

Section snippets

Characteristics of scientific theories

What could a scientific theory of consciousness tell us, and what would be the limits of such a theory? The physical sciences are often taken as the paradigm for rigorous scientific theory. In this domain theories have achieved a high degree of mathematical sophistication over the last 400 years. To define the nature of an effective theory of consciousness it is useful to analyze the nature of theories in the physical sciences. The aim of our discussion is to give a relatively simple view of

Mechanistic views of consciousness

Such a theory would be a mechanistic explanation, i.e., an explanation which “treats [the cognitive] systems as producing a certain behaviour in a manner analogous to that of machines developed through human technology” (Bechtel & Richardson, 1993). What can be expected of such a theory of consciousness is an appreciation of how causal relationships at a deeper level give rise to the phenomena of consciousness at the phenomenological level, a better understanding of causal relationships at the

The definition of consciousness

Even the concept of consciousness is controversial, because of the wide range of different phenomena to which the term “conscious” is applied, and because of the difficulty of objective measurement. A recent attempt at a definition was that of Block (1995) who distinguished between access consciousness, monitoring consciousness, self consciousness, and phenomenal consciousness. Access conscious was defined as the ability to report and act upon experiences. Block suggested that this ability is

The function of consciousness

Another question which must be addressed is the functional or behavioural value of consciousness. One suggestion is the veto view of Libet (1985), or a more general counterbalance view (Kelley & Jacoby, 1993). However, in these views, the reason unconsciously initiated actions need counterbalance, whether in the form of occasional veto or some other form, is not addressed (Sun, 1999). Reber (1989), Stanley, Mathews, Buss, and Kotler-Cope (1989), and others have proposed that conscious and

The psychology of consciousness

There has been extensive experimental work in the areas of perception, memory and skill acquisition (see for example Kirsner et al., 1998). This work has often been interpreted in terms of theoretical distinctions between implicit and explicit mental processes and between declarative and procedural knowledge, There has been criticism of these distinctions, but we believe that the level of description immediately below the general definition of consciousness and its functions should describe

Architectural models of consciousness

At the next level of descriptive detail, separation of the cognitive system into subsystems is required. The separation must be such that the causal relationships between processes in the different subsystems correspond with the causal relationships at higher levels. The argument was made earlier that there are reasons for the existence of a functional modular hierarchy in the brain which provides the appropriate separation. In such a hierarchy, information exchange between modules is required

Criteria for an effective architectural model

Given the objective of a mechanistic or scientific theory of consciousness defined as a consistent hierarchy of causal descriptions from cognitive to physiological levels, what are the essential requirements for an architectural model?
Firstly, causal relationships at the definition and psychological levels must have corresponding causal relationships in the model. For example, the relationships that conscious processes can cause complex verbal reports but unconscious processes cannot must

Descriptions of CLARION and RA

As a first step, the architectures of the two models will be described in more detail.

Accounting for causal relationships at the highest level

In 4 The definition of consciousness, 5 The function of consciousness a number of causal relationships were identified at the levels of the definition of consciousness and the functional role of consciousness. This sections evaluates how causal relationships within the models correspond with these high level relationships. In summary these causal relationships are firstly that sensory input can cause a conscious activation, an unconscious activation, or both. Secondly, an unconscious activation

Accounting for causal relationships at more detailed psychological levels

It is not enough to account for a general difference between conscious and unconscious processes. A viable theory must demonstrate the capability to account for detailed psychological causal relationships. To illustrate this process, we will compare the ability of the two models to account for phenomena across the range of psychological processes discussed earlier.

Correspondence with physiology

The issues here are whether a model maps into known physiology. If so, firstly there should be a functional correspondence between the major subsystems of the model and major neural structures such as cortex and subcortical structures. When there are activations in a model subsystem which cause activations in another subsystem there should be a causal relationship between activations in the corresponding neural structures. Secondly, there should be functional correspondence between more

Discussion

It has been argued that a scientific theory of consciousness requires a hierarchy of consistent causal descriptions of phenomena on many levels from the level of phenomenal consciousness to psychological and physiological levels, in which the deeper levels have lower information density but higher descriptive complexity. A consistent description at one level alone is necessary but not sufficient to demonstrate theoretical viability. It has been further argued that biological brains are likely

Conclusions

Firstly, a scientific theory of consciousness requires construction of a hierarchy of consistent causal descriptions from physiology through a series of intermediate levels to conscious phenomena. It is inadequate to only look for neural correlates of consciousness or to model cognitive data without reference to physiological plausibility or phenomenological analysis. Secondly, although entire conscious processes could in principle be described end-to-end in detail in terms of the activities of

Acknowledgments

Ron Sun acknowledges the support provided in part by ONR Grant N00014-95-1-0440 and ARI Grant DASW01-00-K-0012.

References (103)

  • J.R. Anderson

    Rules of the mind

    (1993)
  • B. Baars

    A cognitive theory of consciousness

    (1988)
  • B. Baars

    In the theatre of consciousness

    Journal of Consciousness Studies

    (1997)
  • L. Bainbridge

    Verbal reports as evidence of the process operator’s knowledge

    International Journal of Man–Machine Studies

    (1977)
  • Bapi, R., & Doya, K. (1999). MFM: Multiple forward model architecture for sequence learning. In L. Giles, R. Sun...
  • L. Bass et al.

    Software architecture in practice

    (1998)
  • W. Bechtel et al.

    Discovering complexity: Decomposition and localization as strategies in scientific research

    (1993)
  • D.C. Berry

    The problem of implicit knowledge

    Expert Systems

    (1987)
  • D. Berry et al.

    Interactive tasks and the implicit–explicit distinction

    British Journal of Psychology

    (1988)
  • N. Block

    On a confusion about a function of consciousness

    Brain and Behavioral Sciences

    (1995)
  • D. Braddon-Mitchell et al.

    Philosophy of mind and cognition

    (1996)
  • N. Butters

    Alcoholic Korsakoff’s syndrome: An update

    Seminars in Neurology

    (1984)
  • W.H. Calvin

    Cortical columns, modules, and Hebbian cell assemblies

  • P. Carruthers

    The cognitive functions of language

    Behavioural and Brain Sciences

    (2002)
  • D.J. Chalmers

    Facing up to the problem of consciousness

    Journal of Consciousness Studies

    (1995)
  • A. Clark

    The presence of a symbol

    Connection Science

    (1992)
  • N.J. Cohen et al.

    Different memory systems underlying acquisition of procedural and declarative knowledge

  • A. Collirs et al.

    Spreading activation theory of semantic processing

    Psychological Review

    (1975)
  • L.A. Coward

    Pattern thinking

    (1990)
  • L.A. Coward

    A physiologically based theory of consciousness

  • L.A. Coward

    A functional architecture approach to neural systems

    International Journal of Systems Research and Information Systems

    (2000)
  • F. Crick et al.

    Toward a neurobiological theory of consciousness

    Seminars in the Neuroscience

    (1990)
  • A. Damasio

    Neural regionalization of knowledge access

  • A. Damasio

    Decartes’ error

    (1994)
  • Dennett, D. (1991). Consciousness explained. Little...
  • H. Dreyfus et al.

    Mind over machine: The power of human intuition

    (1987)
  • J. Dunn

    Implicit memory and amnesia

  • G. Edelman

    The remembered present: A biological theory of consciousness

    (1989)
  • J.A. Fodor

    Modularity of mind

    (1983)
  • G.B. Forbach et al.

    Repetition and practice effects in a lexical decision task

    Memory & Cognition

    (1974)
  • S. Franklin et al.

    A software agent model of consciousness

    Consciousness and Cognition

    (1999)
  • W. Freeman

    Societies of brains

    (1995)
  • J.A. Goguen

    Complexity of hierarchically organized systems and the structure of musical experiences

    International Journal of General Systems

    (1977)
  • G. Greene

    The elegant universe

    (1999)
  • R. Hadley

    The explicit–implicit distinction

    Minds and Machines

    (1995)
  • Heidegger, M. (1927). Being and time. English translation published by Harper and Row, New York,...
  • L. Hirschfeld et al.

    Mapping the mind: Domain specificity in cognition and culture

    (1994)
  • J. Houk et al.

    A model of how the basal ganglia generate and use neural signals that predict reinforcement

  • E. Hunt et al.

    Unified model of attention and problem solving

    Psychological Review

    (1986)
  • R. Jackendoff

    Consciousness and the computational mind

    (1987)

    • Cited by (16)

    • Metacognition for a Common Model of Cognition

      2018, Procedia Computer Science

    • Memory systems within a cognitive architecture

      2012, New Ideas in Psychology
      Citation Excerpt :

      The more information a cognitive agent encodes, the more difficult it is to devise a system that can deliver the right information at the right moment (and do so without overwhelming it with less relevant or irrelevant materials; Klein et al., 2002). Modularization of memory such as the division above (see more discussions below) is therefore of importance (see Coward & Sun, 2004 regarding issues of modularization). Similarly, different informational needs of an agent may require different memory processes and different memory characteristics, and therefore likely different memory modules/systems.

    • Theoretical status of computational cognitive modeling

      2009, Cognitive Systems Research
      Citation Excerpt :

      In contrast to this higher descriptive (Kolmogorov) complexity of phenomena, the description of a theory per se at a deeper level often has a lower Kolmogorov complexity (although not always). This often means that the number of different types of entities that are needed at a deeper level is often smaller than that at a higher level, and the number of different types of relationships at a deeper level is often smaller as well (Coward & Sun, 2004). Thus, theories at a deeper level may often be more “succinct”.

    • LIDA: A systems-level architecture for cognition, emotion, and learning

      2014, IEEE Transactions on Autonomous Mental Development

    • Towards an integrative theory of consciousness: Part 2 (An anthology of various other models)

      2013, Mens Sana Monographs

    • Prolegomena to integrating cognitive modeling and social simulation

      2005, Cognition and Multi Agent Interaction from Cognitive Modeling to Social Simulation
    View all citing articles on Scopus
    View full text
    Copyright © 2003 Elsevier Inc. All rights reserved.