Elsevier

Neural Networks

Volume 20, Issue 9, November 2007, Pages 1032-1039
Neural Networks

2007 Special Issue
Consciousness and metarepresentation: A computational sketch

https://doi.org/10.1016/j.neunet.2007.09.011Get rights and content

Abstract

When one is conscious of something, one is also conscious that one is conscious. Higher-Order Thought Theory [Rosenthal, D. (1997). A theory of consciousness. In N. Block, O. Flanagan, & G. Güzeldere (Eds.), The nature of consciousness: Philosophical debates. Cambridge, MA: MIT Press] takes it that it is in virtue of the fact that one is conscious of being conscious, that one is conscious. Here, we ask what the computational mechanisms may be that implement this intuition. Our starting point is Clark and Karmiloff-Smith’s [Clark, A., & Karmiloff-Smith, A. (1993). The cognizer’s innards: A psychological and philosophical perspective on the development of thought. Mind and Language, 8, 487–519] point that knowledge acquired by a connectionist network always remains “knowledge in the network rather than knowledge for the network”. That is, while connectionist networks may become exquisitely sensitive to regularities contained in their input–output environment, they never exhibit the ability to access and manipulate this knowledge as knowledge: The knowledge can only be expressed through performing the task upon which the network was trained; it remains forever embedded in the causal pathways that developed as a result of training. To address this issue, we present simulations in which two networks interact. The states of a first-order network trained to perform a simple categorization task become input to a second-order network trained either as an encoder or on another categorization task. Thus, the second-order network “observes” the states of the first-order network and has, in the first case, to reproduce these states on its output units, and in the second case, to use the states as cues in order to solve the secondary task. This implements a limited form of metarepresentation, to the extent that the second-order network’s internal representations become re-representations of the first-order network’s internal states. We conclude that this mechanism provides the beginnings of a computational mechanism to account for mental attitudes, that is, an understanding by a cognitive system of the manner in which its first-order knowledge is held (belief, hope, fear, etc.). Consciousness, in this light, thus involves knowledge of the geography of one own’s internal representations — a geography that is itself learned over time as a result of an agent’s attributing value to the various experiences it enjoys through interaction with itself, the world, and others.

Section snippets

The radical plasticity thesis

We would thus like to defend the following claim: Conscious experience occurs if and only if an information processing system has learned about its own representations of the world. To put this claim even more provocatively: Consciousness is the brain’s theory about itself, gained through experience interacting with the world, others, and, crucially, with itself. We call this claim the “Radical Plasticity Thesis”, for its core is the notion that learning is what makes us conscious. How so? The

Simulations: The digits problem

We illustrate two ways in which metarepresentations can be operationalized and what this might teach us about consciousness. Both simulations involve a first-order network that has to perform a simple task such as digit recognition, and a higher-order network that “observes” the internal states of the first-order network. This second network is thus wholly independent from the causal chain set up by the first-order network.

In the first simulation the higher-order network is simply trained to

Metarepresentation

The simulations sketched above illustrate how a network can be trained to observe the internal states of another network in such a manner that it can use this information to perform tasks that require knowledge of the structure of these internal states — either to reconstruct the corresponding inputs and outputs, or to actually evaluate the extent to which these internal representations will result in successful performance. In both cases, it is interesting to note that while the higher-order,

Conclusion

Thus we end with the following idea, which is the heart of the “Radical Plasticity Thesis”: The brain continuously and unconsciously learns not only about the external world, but about its own representations of it. The result of this unconscious learning is conscious experience, in virtue of the fact that each representational state is now accompanied by (unconsciously learnt) metarepresentations that convey the mental attitude with which these first-order representations are held. Thus, from

Acknowledgements

A.C. is a Research Director with the National Fund for Scientific Research (F.R.S.-FNRS, Belgium). B.T. is a postdoctoral fellow at the ULB — Université Libre de Bruxelles. A.P. is a Research Fellow of the Fonds National de la Recherche Scientifique (F.R.S-FNRS, Belgium). This work was supported by an institutional grant from the Université Libre de Bruxelles to A.C., by Concerted Research Action 06/11-342 titled “Culturally modified organisms: What it means to be human in the age of culture”,

References (37)

  • D. Bierman et al.

    Intuitive decision making in complex situations: Somatic markers in an artificial grammar learning task

    Cognitive, Affective & Behavioral Neuroscience

    (2005)
  • D.J. Chalmers

    The conscious mind: In search of a fundamental theory

    (1996)
  • D.J. Chalmers

    The hard problem of consciousness

  • A. Clark et al.

    The cognizer’s innards: A psychological and philosophical perspective on the development of thought

    Mind and Language

    (1993)
  • A. Cleeremans

    Conscious and unconscious cognition: A graded, dynamic perspective

  • S. Dehaene et al.

    A neuronal model of a global workspace in effortful cognitive tasks

    Proceedings of the National Academy of Sciences of the United States of America

    (1998)
  • D.C. Dennett

    Consciousness explained

    (1991)
  • Z. Dienes

    Assumptions of subjective measures of unconscious mental states: Higher order thoughts and bias

    Journal of Consciousness Studies

    (2004)
  • Cited by (72)

    • Imagination: The dawn of consciousness: Fighting some misconceptions in the discussion about consciousness

      2023, Physiology and Behavior
      Citation Excerpt :

      Some ToCs merely aim at defining consciousness; though the definitions extensively differ, they tend to define it as a property, a function or a purpose, and such properties, functions and purposes are reasonably consistent. Some examples are: “Consciousness may be thought as the global result of integrative processes taking place at various levels of miniaturization in plastic mosaics” [1]; “Consciousness is the brain's theory about itself” [12]; “Consciousness is the result of all information processing activities that occur in the mind; it is the result of all processes that occur in the brain” [54]; Gelepithis [32] formulated a theory that tries to rigorously define consciousness but ends up being essentially descriptive: what is needed for a stimulus in its context to be the object of conscious attention. Conversely, other ToCs aim at describing how consciousness comes about; in this case, three completely distinct approaches can be identified.

    • Towards characterizing the canonical computations generating phenomenal experience

      2022, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      This is not a paper in which I will argue about whether confidence and consciousness are identical, so to answer this question in a word: no. Confidence and conscious awareness are certainly not the same thing (Rosenthal, 2000, 2019; Sherman et al., 2015). It is true that many links have been drawn between confidence and consciousness, highlighting both their similarities and discrepancies; I do not intend to review this discussion here, as it has been elegantly and comprehensively described elsewhere (e.g., (Cleeremans et al., 2007; Pasquali et al., 2010; Rausch and Zehetleitner, 2016; Rosenthal, 2019; Sandberg et al., 2010a, 2010b; Wierzchoń et al., 2014; Zehetleitner and Rausch, 2013), and most recently (Morales and Lau, 2021), among many others). For the purposes of the argument presented here, however, suffice it to say that while consciousness and confidence are likely intimately related (Fleming, 2020; Y. Ko and Lau, 2012; Rausch et al., 2017; Rausch and Zehetleitner, 2016, 2018; Rausch et al., 2019), most researchers take the position that they are not equivalent (Morales and Lau, 2021; Rosenthal, 2019).

    • The Bayesian brain: An evolutionary approach to cognition

      2021, Encyclopedia of Behavioral Neuroscience: Second Edition
    • Learning to Be Conscious

      2020, Trends in Cognitive Sciences
    View all citing articles on Scopus
    View full text