A model of the hierarchy of behaviour, cognition, and consciousness

https://doi.org/10.1016/j.concog.2005.04.008Get rights and content

Abstract

Processes comparable in important respects to those underlying human conscious and non-conscious processing can be identified in a range of species and it is argued that these reflect evolutionary precursors of the human processes. A distinction is drawn between two types of processing: (1) stimulus-based and (2) higher-order. For ‘higher-order,’ in humans the operations of processing are themselves associated with conscious awareness. Conscious awareness sets the context for stimulus-based processing and its end-point is accessible to conscious awareness. However, the mechanics of the translation between stimulus and response proceeds without conscious control. The paper argues that higher-order processing is an evolutionary addition to stimulus-based processing. The model’s value is shown for gaining insight into a range of phenomena and their link with consciousness. These include brain damage, learning, memory, development, vision, emotion, motor control, reasoning, the voluntary versus involuntary debate, and mental disorder.

Introduction

This paper: (a) examines some of the characteristics of processes, which, in humans, are described as either ‘conscious’ or ‘unconscious,’ (b) links this to their biological embodiment, and (c) speculates on their evolutionary history. To achieve this, it theorizes about the processes’ evolutionary origins: it will show where certain ‘precursor processes,’ which are evident in a range of species, exhibit a similar distinction in performance to human non-conscious and conscious processes. That is, it will argue for evolutionary continuity. The paper will also look at interactions between conscious and non-conscious processing, in the context of the similar processes possessed by a range of species. As a framework, a model of behaviour and cognition (Toates, 1995, Toates, 1998, Toates, 2000, Toates, 2001, Toates, 2002a, Toates, 2002b, Toates, 2004, Toates, 2005) will be extended to the study of consciousness. It is suggested that researchers could benefit from a greater appreciation of design solutions employed more widely across species.

The paper will investigate how a system of multiple layers of interacting control can describe features of behaviour, cognition, and consciousness. As a start, it assumes that the simplest and evolutionarily oldest solutions to producing adaptive behaviour involve neither sophisticated cognition nor consciousness (cf. Reber, 1997). The paper then suggests what are the strengths and weaknesses of such relatively simple ‘stimulus-based’ solutions. It identifies the adaptive value of the evolutionary emergence of additional and higher-order processes. Such processes are commonly seen to involve representations of the environment, as implied by such terms as ‘cognitive map’ and ‘expectation.’ There is controversy on whether the term ‘representation’ really is needed to account for behaviour (Clark, 1998, Clark, 1999, Van Gelder, 1998). Therefore, the paper will largely avoid this issue by simply using the term ‘higher order.’ The higher-order processes are defined by their flexibility and also by exclusion, as those that are evolutionarily more recent and which do not mediate behaviour in a direct stimulus-based way. However, the fact that terms such as expectancy, error, model, disparity, and cognitive map will be used in various places implies a belief in something resembling a traditional representation, even though the strength of the model does not rest or fall with the validity of this notion.

Of course, the question of animal awareness (‘phenomenal consciousness’) is not a new one and we do not know whether non-humans experience phenomenal consciousness. Others have speculated about animal awareness on the basis of behaviour, brain mechanisms and evolutionary considerations (Dawkins, 1990, Griffin, 1981). However, the present paper adopts a somewhat different perspective from these authors by pursuing a contrast between processes that appear to map onto the human conscious/unconscious divide. We know that many species exhibit higher-order cognition of a kind that would be open to conscious introspection in humans, e.g., goal-directedness, flexible use of memory, extrapolation to beyond sensory stimulation, and resisting habitual behaviour. We can also identify the brain regions underlying these cognitive capacities. It is argued that a study of the bases of such information processing in non-humans is relevant to how, in humans, features of consciousness are associated with similar cognition and underlying brain regions.

These ideas represent a form of hierarchical control and this notion will provide a framework throughout. The hierarchical model conforms to evidence suggesting that new structures emerge in evolution by building onto existing structures, so-called ‘tinkering’ (Jacob, 1977, Rozin, 1976a). More recently evolved brain processes co-exist with, and exert modulation over, evolutionarily older processes (Panksepp, 1998, Rinn, 1984). Reber (1997) argued that consciousness is an evolutionary stage that builds upon older unconscious cognition.

As a first approximation, the model will be based on the following distinction between processes. There are stimulus-based controls of behaviour, which are assumed to be evolutionarily old and appear to correspond to Reber’s term ‘unconscious cognition.’ They provide rather direct prescriptions for behaviour triggered by raw stimulus input. Where memory is involved in their expression, such on-line processes exploit what is termed implicit or procedural memory (Squire, 1986).

There are also more evolutionarily recent higher-order controls that correspond in some ways to Reber’s term ‘conscious cognition.’ They exploit what in humans is termed explicit or declarative memory. The higher-order controls appear to be an evolutionary development that builds upon, and co-exists with, older more mechanistic stimulus-based controls. Certain ‘design requirements’ associated with stimulus-based and higher-order control will be identified. It will be shown where higher-order information processing is intrinsically associated with human consciousness.

It will be argued that in the case of humans (and possibly some other species) conscious processing sets the context in which the stimulus-based process operates and the product of this processing can gain access to consciousness. For example, we set goals that facilitate stimulus-based processing that is compatible with the goal. However, we do not have conscious access to any of the stages of such processing. Under some conditions, conscious controls can also operate to direct behaviour even though this goes against the tendency arising from stimuli.

In discussions of consciousness, there is frequently a parallel consideration of the voluntary versus involuntary nature of different behavioural controls (Baars, 1988). In humans, this is sometimes framed in terms of ‘intentional’ versus ‘automatic’ controls and, in more philosophical discourse, as free-will versus determinism. Though this topic is somewhat beyond our brief, the ideas expressed here are relevant to it and so it will be addressed later. The paper will not concern itself with the ‘hard’ problem of how phenomenal consciousness arises (Chalmers, 1996). Rather, it will consider only a range of ‘easy’ problems concerning the kind of information processing and control of action that is described as either conscious or unconscious.

Of course, the reader might feel that it is almost tautological to claim that animals are moved by either stimuli or higher-order controls or both, for what else could there be? However, the article is designed to show how these controls interact, the functional significance of their interaction, and where this can provide a new source of insight and synthesis relevant to consciousness.

The paper discusses first the likely design considerations underlying the evolutionary emergence of stimulus-based and higher-order controls of behaviour.

Section snippets

External stimuli

Some features of behaviour (e.g., behaviour described by the terms ‘reflex’ and ‘modal action pattern’) depend upon the fundamental design principle that organisms are constructed so as to react to stimuli (Barlow, 1977, Braitenberg, 1984, Gallistel, 1980, Goodale and Milner, 2003, Rinn, 1984, Schneirla, 1959, Sherrington, 1948, Tulving, 1985). Survival requires that certain key physical stimuli such as nutrients trigger approach, whereas escape is triggered from excesses of, amongst other

Bases

The term ‘higher-order’ refers to a system that is disengaged relative to the direct stimulus–response links. It permits such things as cognitive maps to be exploited. In either rudimentary or more sophisticated forms, such control is evident in both vertebrates and certain invertebrates, e.g., the digger wasp (Braitenberg, 1984, Gallistel, 1980). At least some version of higher-order control therefore appears to have been invented more than once in evolution. Since, by definition, within this

A dichotomy of processes but not of control

As a first approximation, two distinct types of process have just been described. The evidence (to be reviewed in this and later sections), from humans and non-humans, points to dual control of behaviour by stimulus-based and higher-order controls acting in concert. The model suggests that, in the competition for behavioural control, stimuli exert an influence in proportion to the modulated strength of the signal that they trigger in the CNS. In this way, a powerful stimulus can occasionally

Background

The model fits a tradition of ‘hierarchical control,’ which describes different layers of control. Hierarchical theories and models of behaviour have a distinguished history in psychology (Freud, 1969, Gallistel, 1980, MacLean, 1990, Panksepp, 1998, Powers, 1973, Roitblat, 1991, Schore, 2001), ethology (Baerends, 1976, Tinbergen, 1969), and the neurosciences (Hughlings Jackson, see Rinn, 1984, Taylor, 1958). Some address the phenomenon of consciousness (Johnson-Laird, 1988, Norman and Shallice,

Changes in weight of controls

Stimulus-based and higher-order controls are involved in any instance of behaviour. However, the evidence suggests that their relative weight (‘efficacy as a control factor’) can change with the following:

  • (a)

    Time within the life-span. Across various species, a similar shift of weight occurs with age. In dogs learning an instrumental task, there is a shift towards a more automatic (‘stimulus–response’) mode of control with increased age (Chan et al., 2002). In humans, there is a change in the power

Attention and layers of control

This section briefly relates the phenomena of attention to the proposed model.

A move to stimulus-based control

With extensive repetition of flexible behaviour, a behavioural system can change its parameters to acquire some features of a reflex (Dickinson, 1985) or modal action pattern (Barlow, 1977). Behaviour becomes automatic, stereotyped, and under stimulus control, i.e., where an automatic process will suffice the system adjusts to become this. The adjustment points to the adaptive value of exploiting the benefits of speed and simplicity, captured by the term ‘automaticity’ (Baars, 1988, Schneider

Basic principles

As indicated at various points already, whether taking the perspective of learning or memory, evidence points to stimulus-based and higher-order layers of control. Further development of this topic is presented in Section 11 where it is related to brain structure.

Motor control

Willingham (1998) presented a hierarchical model of motor control and skill learning. At the top, is a strategic process, involving goal-setting and biologically rooted in the dorsolateral frontal cortex. The performance of this layer is associated with conscious awareness. The high layer uses an allocentric representation of the world and the actor can articulate what (s)he is trying to do to change the world. At lower levels in the hierarchy, e.g., basal ganglia and spinal cord,

Basic principles

Vision and the control of eye movements show evidence of layers of control with different evolutionary histories but with shared responsibility. Paillard (1987) proposed a layer of control that (p. 43):

…allows neural processing to step back from the immediate sensory input and to become progressively free from the environmental constraints under which sensorimotor analyzers have to work.

In humans, this maps onto older (non-conscious) stimulus-based controls and evolutionarily newer processing (

Motivation, rationality, stimuli, and higher-order controls

Philosophical discourse is concerned with the ability of humans (or lack of it) to make rational (‘conscious’) decisions. A feature of decision-making appears to be a capacity to weigh the probable negative long-term future outcome of an action against the immediate positive gains by employing somatic markers (Adolphs, 1999, Damasio, 1996), i.e., a capacity to weigh appropriately stimuli and higher-order determinants is indicated. Evidence points to a disruption of this capacity following

The basic idea

A notion passionately advanced and criticized at one time was that ontogeny recapitulates (‘recaptures’) phylogeny. That is, changes over evolutionary time can be observed in speeded-up form over development. This idea fell into disfavour, though it then saw a qualified revival (Gould, 1977; cf. Bjorklund and Pellegrini, 2002, Rozin, 1976b). Evidence points to one respect in which the principle is valid: in both ontogeny and phylogeny, there emerges an increasing degree of higher-order control

Pathology of thought and behaviour

Insight can sometimes be gained by observing how brain and behavioural systems can ‘go wrong.’ Space precludes a detailed discussion but this section attempts to give a flavour of the argument.

General

A range of evidence, human and non-human, points to a dichotomy in types of control that is manifest in their shared responsibility for behaviour and cognition. There can be a shift of weight between types of control and competition between them. What appears to be an adaptive (‘dual’) solution for the control of behaviour can sometimes be reinvented by evolution. This is revealed in: (i) vision in the dichotomy between subcortical and cortical being reinvented within cortical processing and

Acknowledgments

I am most grateful to Bernard Baars, Kent Berridge, Daniel Nettle, Jaak Panksepp, and two anonymous referees for their valuable comments on the paper.

References (218)

  • N.I. Eisenberger et al.

    Why rejection hurts: A common neural alarm system for physical and social pain

    Trends in Cognitive Sciences

    (2004)
  • J.S.B.T. Evans

    In two minds: Dual-process accounts of reasoning

    Trends in Cognitive Sciences

    (2003)
  • J.J. Furedy et al.

    Human Pavlovian HR decelerative conditioning with negative tilt as US: A review of some S–R, stimulus-substitution evidence

    International Journal of Psychophysiology

    (1989)
  • H. Gleitman et al.

    Learning and memory

  • V. Goel et al.

    Explaining modulation of reasoning by belief

    Cognition

    (2003)
  • A.M. Graybiel

    The basal ganglia and chunking of action repertoires

    Neurobiology of Learning and Memory

    (1998)
  • R. Hirsh

    The hippocampus and contextual retrieval of information from memory: A theory

    Behavioural Biology

    (1974)
  • A. Allport

    Visual attention

  • A. Amsel

    Précis of “Frustration Theory: An analysis of dispositional learning and memory”

    Psychonomic Bulletin and Review

    (1994)
  • N.C. Andreasen

    Linking mind and brain in the study of mental illnesses: A project for a scientific psychopathology

    Science

    (1997)
  • N.C. Andreasen

    A unitary model of schizophrenia

    Archives of General Psychiatry

    (1999)
  • J. Atkinson

    The developing visual brain

    (2000)
  • B.J. Baars

    A cognitive theory of consciousness

    (1988)
  • B.J. Baars

    In the theatre of consciousness

    (1997)
  • B.J. Baars et al.

    Consciousness creates access: Conscious goal images recruit unconscious action routines, but goal competition serves to “liberate” such routines, causing predictable slips

  • A. Baddeley

    Amnesia: A minimal model and an interpretation

  • J.A. Bargh et al.

    Beyond behaviourism: On the automaticity of higher mental processes

    Psychological Bulletin

    (2000)
  • G.W. Barlow

    Modal action patterns

  • A.T. Beck

    Cognitive therapy and the emotional disorders

    (1976)
  • M.E. Bitterman

    The comparative analysis of learning

    Science

    (1975)
  • M.E. Bitterman

    Evidence of divergence in vertebrate learning

    Behavioral and Brain Sciences

    (1987)
  • D.F. Bjorklund et al.

    The origins of human nature

    (2002)
  • J.E. Bogen

    The callosal syndromes

  • V. Braitenberg

    Vehicles: Experiments in synthetic psychology

    (1984)
  • K. Breland et al.

    The misbehaviour of organisms

    American Psychologist

    (1961)
  • C.R. Brewin

    Theoretical foundations of cognitive-behaviour therapy for anxiety and depression

    Annual Review of Psychology

    (1996)
  • B. Bridgeman

    Conscious vs unconscious processes

    Theory and Psychology

    (1992)
  • G. Bronson

    The postnatal growth of visual capacity

    Child Development

    (1974)
  • A.D.F. Chan et al.

    Visuospatial impairments in aged canines (Canis familiaris): The role of cognitive-behavioural flexibility

    Behavioural Neuroscience

    (2002)
  • D.J. Chalmers

    The conscious mind: In search of a fundamental theory

    (1996)
  • M.E. Claparède

    Récognition et Moité

    Archives de Psychologie

    (1911)
  • A. Clark

    Where brain, body and world collide

    Daedalus

    (1998)
  • A. Clark

    Visual awareness and visuomotor action

    Journal of Consciousness Studies

    (1999)
  • R.E. Clark et al.

    Classical conditioning and brain systems: The role of awareness

    Science

    (1998)
  • G. Claxton

    Whodunnit? Unpicking the ‘seems’ of free will

    Journal of Consciousness Studies

    (1999)
  • J.D. Cohen et al.

    Context, cortex, and dopamine: A connectionist approach to behaviour and biology in schizophrenia

    Psychological Review

    (1992)
  • M. Corbetta et al.

    Control of goal-directed and stimulus-driven attention in the brain

    Nature Reviews Neuroscience

    (2002)
  • L.P. Crespi

    Quantitative variation of incentive and performance in the white rat

    The American Journal of Psychology

    (1942)
  • A.R. Damasio

    The somatic marker hypothesis and the possible functions of the prefrontal cortex

    Philosophical Transactions of the Royal Society of London B

    (1996)
  • A.R. Damasio et al.

    Emotional disturbances associated with focal lesions of the limbic frontal lobe

  • Cited by (89)

    • Modeling the dynamics of action

      2021, Measuring and Modeling Persons and Situations
    • The history of dynamic approaches to personality

      2021, The Handbook of Personality Dynamics and Processes
    • Neural signs and mechanisms of consciousness: Is there a potential convergence of theories of consciousness in sight?

      2020, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      50 years ago, consciousness was basically not present in neuroscience as it was conceived a subject of philosophy (rather than science). 30 years ago consciousness started to enter neuroscience where, initially, it was still considered a matter of (feedforward) convergence of low level towards high level processing, towards some ‘pinnacle’ of processing where consciousness would reside (Toates, 2006) as it, to some extent, re-surfaces in the HOT with its assumption of some specialized function called ‘higher-order thought’ (Brown et al., 2019; Lau and Rosenthal, 2011). ( Fig. 4a).

    • Six clarifications for behaviour systems

      2020, Behavioural Processes
      Citation Excerpt :

      The lesson on this point is to remember that network diagrams depicting behaviour systems can appear similar, and yet carry drastically different meaning. Hierarchical models are drawn with a wide variety of terms and a wide variety of purposes (e.g. compare the models treated above with Toates, 2006; Balleine et al., 2015; Badcock et al., 2019; Ramstead et al., 2019). Behaviour systems modellers should remember that they have this freedom to choose terms to fit a given purpose, but also that this freedom obliges attention to a given modeller’s stated intentions, and to resist the temptation to make guesses from a glance at the pictures.

    View all citing articles on Scopus
    View full text