CoChat

Sentience and the Origins of Consciousness: From Cartesian Duality to Markovian Monism

Open access

The DOI matches its Crossref record; 145 of the 145 references it cites were found in a registry or an index.

Abstract

This essay addresses Cartesian duality and how its implicit dialectic might be repaired using physics and information theory. Our agenda is to describe a key distinction in the physical sciences that may provide a foundation for the distinction between mind and matter, and between sentient and intentional systems. From this perspective, it becomes tenable to talk about the physics of sentience and ‘forces’ that underwrite our beliefs (in the sense of probability distributions represented by our internal states), which may ground our mental states and consciousness. We will refer to this view as Markovian monism, which entails two claims: (1) fundamentally, there is only one type of thing and only one type of irreducible property (hence monism). (2) All systems possessing a Markov blanket have properties that are relevant for understanding the mind and consciousness: if such systems have mental properties, then they have them partly by virtue of possessing a Markov blanket (hence Markovian). Markovian monism rests upon the information geometry of random dynamic systems. In brief, the information geometry induced in any system—whose internal states can be distinguished from external states—must acquire a dual aspect. This dual aspect concerns the (intrinsic) information geometry of the probabilistic evolution of internal states and a separate (extrinsic) information geometry of probabilistic beliefs about external states that are parameterised by internal states. We call these intrinsic (i.e., mechanical, or state-based) and extrinsic (i.e., Markovian, or belief-based) information geometries, respectively. Although these mathematical notions may sound complicated, they are fairly straightforward to handle, and may offer a means through which to frame the origins of consciousness.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3390/e22050516

Cites

145 of 145 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 3 (Markov blankets and other formulations): This schematic illustrates the various interpretations of a gradient flow on surprisal. Recall that the existence of a Markov blanket implies a certain lack of influences among internal, blanket, and external states. At nonequilibrium steady-state, these independencies have an important consequence; internal and active states are the only states that are not influenced by external states, which means their dynamics (i.e., perception and action) are a function of, and only of, particular states; i.e., a particular surprisal.8 This surprisal has a number of interesting interpretations. Given it is the negative log probability of finding a particle or creature in a particular state, minimising particular surprisal corresponds to maximising the value of a particle’s state. This interpretation is licensed by the fact that the states with a high probability are, by definition, attracting states. On this view, one can then spin-off an interpretation in terms of reinforcement learning [30], optimal control theory [31] and, in economics, expected utility theory [39]. Indeed, any scheme predicated on the optimisation of some objective function can now be cast in terms of minimising a particular surprisal—in terms of perception and action (i.e., the flow of internal and active states). The minimisation of particular surprisal leads to a series of influential accounts of neuronal dynamics; including the principle of maximum mutual information [40,41], the principles of minimum redundancy and maximum efficiency [33] and—as we will see later—the free energy principle [42]. Crucially, the average or expected surprisal (over time or particular states of being) corresponds to entropy. This means that action and perception look as if they are minimising a particular entropy. The implicit resistance to the second law of thermodynamics leads us to theories of self-organisation, such as synergetics in physics [29,43,44] or homoeostasis in physiology [35,45,46]. Finally, the probability of any particular states given a Markov blanket (m) is, on a statistical view, model evidence [18,47]. This means that all the above formulations are internally consistent with things like the Bayesian brain hypothesis, evidence accumulation and predictive coding; most of which inherit from Helmholtz’s motion of unconscious inference [48], later unpacked in terms of perception as hypothesis testing in 20th century psychology [49] and machine learning [50]. In short, the very existence of something leads in the natural way to a whole series of optimisation frameworks in the physical and life sciences that lends each a construct validity in relation to the others.
Figure 3 (Markov blankets and other formulations): This schematic illustrates the various interpretations of a gradient flow on surprisal. Recall that the existence of a Markov blanket implies a certain lack of influences among internal, blanket, and external states. At nonequilibrium steady-state, these independencies have an important consequence; internal and active states are the only states that are not influenced by external states, which means their dynamics (i.e., perception and action) are a function of, and only of, particular states; i.e., a particular surprisal.8 This surprisal has a number of interesting interpretations. Given it is the negative log probability of finding a particle or creature in a particular state, minimising particular surprisal corresponds to maximising the value of a particle’s state. This interpretation is licensed by the fact that the states with a high probability are, by definition, attracting states. On this view, one can then spin-off an interpretation in terms of reinforcement learning [30], optimal control theory [31] and, in economics, expected utility theory [39]. Indeed, any scheme predicated on the optimisation of some objective function can now be cast in terms of minimising a particular surprisal—in terms of perception and action (i.e., the flow of internal and active states). The minimisation of particular surprisal leads to a series of influential accounts of neuronal dynamics; including the principle of maximum mutual information [40,41], the principles of minimum redundancy and maximum efficiency [33] and—as we will see later—the free energy principle [42]. Crucially, the average or expected surprisal (over time or particular states of being) corresponds to entropy. This means that action and perception look as if they are minimising a particular entropy. The implicit resistance to the second law of thermodynamics leads us to theories of self-organisation, such as synergetics in physics [29,43,44] or homoeostasis in physiology [35,45,46]. Finally, the probability of any particular states given a Markov blanket (m) is, on a statistical view, model evidence [18,47]. This means that all the above formulations are internally consistent with things like the Bayesian brain hypothesis, evidence accumulation and predictive coding; most of which inherit from Helmholtz’s motion of unconscious inference [48], later unpacked in terms of perception as hypothesis testing in 20th century psychology [49] and machine learning [50]. In short, the very existence of something leads in the natural way to a whole series of optimisation frameworks in the physical and life sciences that lends each a construct validity in relation to the others.CC BY · PubMed Central
Figure 1 (Markov blankets): This schematic illustrates the partition of systemic states into internal states (blue) and hidden or external states (cyan) that are separated by a Markov blanket—comprising sensory (magenta) and active states (red). The upper panel shows this partition as it would be applied to action and perception in the brain. The ensuing self-organisation of internal states then corresponds to perception, while action couples brain states back to external states. The lower panel shows the same dependencies but rearranged so that the internal states are associated with the intracellular states of a Bacillus, while the sensory states become the surface states or cell membrane overlying active states (e.g., the actin filaments of the cytoskeleton).
Figure 1 (Markov blankets): This schematic illustrates the partition of systemic states into internal states (blue) and hidden or external states (cyan) that are separated by a Markov blanket—comprising sensory (magenta) and active states (red). The upper panel shows this partition as it would be applied to action and perception in the brain. The ensuing self-organisation of internal states then corresponds to perception, while action couples brain states back to external states. The lower panel shows the same dependencies but rearranged so that the internal states are associated with the intracellular states of a Bacillus, while the sensory states become the surface states or cell membrane overlying active states (e.g., the actin filaments of the cytoskeleton).CC BY · PubMed Central
Figure 2 (density dynamics and pullback attractors): This figure illustrates the fundaments of density or ensemble dynamics in random dynamical systems—of the sort described by the Langevin equation. The left panel pictures some arbitrary random attractor (a.k.a., a pullback attractor) that can be thought of in two ways: first, it can be considered as the trajectory of (two) systemic states as they evolve over time. For example, these two states could be the depolarisation and current of a nerve cell, over several minutes. At a larger timescale, this trajectory could reflect your daily routine, getting up in the morning, having a cup of coffee, going to work and so on. It could also represent the slow fluctuations in two meteorological states over the period of a year. The key aspect of this trajectory is that it will—after itinerant wandering and a sufficient period of time—revisit particular regimes of state space. These states constitute the attracting set or pullback attractor. The second interpretation is of a probability density over the states that the system will be found in, when sampled at random. The evolution of the probability density is described by the Fokker-Planck equation. Crucially, when any system has attained nonequilibrium steady state, we know that this density does not change with time. This affords the solution to the Fokker-Planck equation—a solution that means that there is a lawful relationship between the flow of states at any point in state space and the probability density. This solution expresses the flow in terms of gradients of log density or surprisal and the amplitude of random fluctuations. In turn, the nonequilibrium steady-state solution can always be expressed, via the Helmholtz decomposition, in terms of two orthogonal components. One component is a gradient flow that rebuilds probability gradients in a way that is exactly countered by the dispersion of states due to random fluctuations. The other component is a solenoidal or divergence-free flow that circulates on isoprobability contours. These two components are shown in the schematic on the right, in terms of a curl-free gradient flow—that depends only on the amplitude of random fluctuations Γ– and a divergence-free solenoidal flow—that depends upon an antisymmetric matrix Q. This example shows the flow around the peak of a probability density, with a Gaussian form. Please see [1,25] for details.
Figure 2 (density dynamics and pullback attractors): This figure illustrates the fundaments of density or ensemble dynamics in random dynamical systems—of the sort described by the Langevin equation. The left panel pictures some arbitrary random attractor (a.k.a., a pullback attractor) that can be thought of in two ways: first, it can be considered as the trajectory of (two) systemic states as they evolve over time. For example, these two states could be the depolarisation and current of a nerve cell, over several minutes. At a larger timescale, this trajectory could reflect your daily routine, getting up in the morning, having a cup of coffee, going to work and so on. It could also represent the slow fluctuations in two meteorological states over the period of a year. The key aspect of this trajectory is that it will—after itinerant wandering and a sufficient period of time—revisit particular regimes of state space. These states constitute the attracting set or pullback attractor. The second interpretation is of a probability density over the states that the system will be found in, when sampled at random. The evolution of the probability density is described by the Fokker-Planck equation. Crucially, when any system has attained nonequilibrium steady state, we know that this density does not change with time. This affords the solution to the Fokker-Planck equation—a solution that means that there is a lawful relationship between the flow of states at any point in state space and the probability density. This solution expresses the flow in terms of gradients of log density or surprisal and the amplitude of random fluctuations. In turn, the nonequilibrium steady-state solution can always be expressed, via the Helmholtz decomposition, in terms of two orthogonal components. One component is a gradient flow that rebuilds probability gradients in a way that is exactly countered by the dispersion of states due to random fluctuations. The other component is a solenoidal or divergence-free flow that circulates on isoprobability contours. These two components are shown in the schematic on the right, in terms of a curl-free gradient flow—that depends only on the amplitude of random fluctuations Γ– and a divergence-free solenoidal flow—that depends upon an antisymmetric matrix Q. This example shows the flow around the peak of a probability density, with a Gaussian form. Please see [1,25] for details.CC BY · PubMed Central
Figure 4 (autonomous movement). This figure shows the results of simulating active inference (i.e., writing), in terms of conditional expectations about hidden states of the world, consequent predictions about sensory input and the ensuing behaviour. The autonomous dynamics that underlie this behaviour rest upon prior expectations about states with Lotka-Volterra dynamics (c.f., a central pattern generator): these are the six (arbitrarily) coloured lines in the upper left panel. In this generative model, each state is associated with a location in Euclidean space that attracts the agent’s finger. In effect, the internal states then supply predictions of what sensory states should register, if these prior beliefs were true. Active states try to suppress the ensuing prediction error (i.e., sensory surprisal) by reflexively fulfilling expected changes in angular velocity, through exerting forces on the agent’s joints (not shown). The subsequent movement of the arm is traced out in the lower left panel. This trajectory has been plotted in a moving frame of reference, so that it looks like synthetic handwriting (e.g., a succession of ‘j’ and ‘a’ letters). The lower left panels show the activity of one (the fourth attractor) conditional expectation under ‘action’, and ‘action-observation’. During action, sensory states register both the visual and proprioceptive consequences of movement, while under action observation, only visual sensations are available—as if the agent was watching another agent. The red dots correspond to the time bins during which this state exceeded an amplitude threshold of two arbitrary units. They key thing to note here is that this unit responds preferentially when, and only when, the motor trajectory produces a down-stroke, but not an up-stroke. Please see [94] for further details. Furthermore, with a slight delay, this internal state responds during action and action observation. From a biological perspective, this is interesting because it speaks to an empirical phenomena known as mirror neuron activity [95,96,97].
Figure 4 (autonomous movement). This figure shows the results of simulating active inference (i.e., writing), in terms of conditional expectations about hidden states of the world, consequent predictions about sensory input and the ensuing behaviour. The autonomous dynamics that underlie this behaviour rest upon prior expectations about states with Lotka-Volterra dynamics (c.f., a central pattern generator): these are the six (arbitrarily) coloured lines in the upper left panel. In this generative model, each state is associated with a location in Euclidean space that attracts the agent’s finger. In effect, the internal states then supply predictions of what sensory states should register, if these prior beliefs were true. Active states try to suppress the ensuing prediction error (i.e., sensory surprisal) by reflexively fulfilling expected changes in angular velocity, through exerting forces on the agent’s joints (not shown). The subsequent movement of the arm is traced out in the lower left panel. This trajectory has been plotted in a moving frame of reference, so that it looks like synthetic handwriting (e.g., a succession of ‘j’ and ‘a’ letters). The lower left panels show the activity of one (the fourth attractor) conditional expectation under ‘action’, and ‘action-observation’. During action, sensory states register both the visual and proprioceptive consequences of movement, while under action observation, only visual sensations are available—as if the agent was watching another agent. The red dots correspond to the time bins during which this state exceeded an amplitude threshold of two arbitrary units. They key thing to note here is that this unit responds preferentially when, and only when, the motor trajectory produces a down-stroke, but not an up-stroke. Please see [94] for further details. Furthermore, with a slight delay, this internal state responds during action and action observation. From a biological perspective, this is interesting because it speaks to an empirical phenomena known as mirror neuron activity [95,96,97].CC BY · PubMed Central
Figure 5 (epistemic foraging). This figure shows the results of a simulation in which a face was presented to an agent, whose responses were simulated by selecting active states that minimise expected free energy following an eye movement. The agent had three internal images or hypotheses about the stimuli she might sample (an upright face, and inverted face and a rotated face). The agent was presented with an upright face and her conditional expectations were evaluated over 16 (12 ms.) time bins, until the next saccade was emitted. This was repeated for eight saccades. The ensuing eye movements are shown as red dots at the end of each saccade in the upper row. The corresponding sequence of eye movements is shown in the insert on the upper left, where the red circles correspond roughly to the proportion of the visual image sampled. These saccades are driven by prior beliefs about the direction of gaze based upon the saliency maps in the second row. These saliency maps are the expected free energy as a function of policies; namely, where to look next. Note that these maps change with successive saccades as posterior beliefs about external states, including the stimulus, become progressively more precise or confident. Note also that salience is depleted in locations that were foveated in the previous saccade because these locations no longer have an epistemic affordance (i.e., the ability to reduce uncertainty or expected free energy). This is a nice illustration of a ubiquitous phenomenon, known as inhibition of return. Oculomotor responses are shown in the third row in terms of the two hidden oculomotor states, corresponding to vertical and horizontal eye movements. The associated portions of the image sampled (at the end of each saccade) are shown in the fourth row. The final two rows show the posterior beliefs in terms of their sufficient statistics and the stimulus categories, respectively. The posterior beliefs are plotted here in terms of conditional expectations and the 90% confidence interval about the true stimulus. The key thing to note here is that the expectation about the true stimulus supervenes over its competing expectations and, as a result, conditional confidence about the stimulus category increases (the confidence intervals shrink to the expectation). This illustrates the nature of evidence accumulation when selecting a hypothesis or percept the best explains sensory data. Within saccade accumulation is evident even during the initial fixation with further stepwise decreases in uncertainty as salient information is sampled at successive saccades. Please see [98] for further details.
Figure 5 (epistemic foraging). This figure shows the results of a simulation in which a face was presented to an agent, whose responses were simulated by selecting active states that minimise expected free energy following an eye movement. The agent had three internal images or hypotheses about the stimuli she might sample (an upright face, and inverted face and a rotated face). The agent was presented with an upright face and her conditional expectations were evaluated over 16 (12 ms.) time bins, until the next saccade was emitted. This was repeated for eight saccades. The ensuing eye movements are shown as red dots at the end of each saccade in the upper row. The corresponding sequence of eye movements is shown in the insert on the upper left, where the red circles correspond roughly to the proportion of the visual image sampled. These saccades are driven by prior beliefs about the direction of gaze based upon the saliency maps in the second row. These saliency maps are the expected free energy as a function of policies; namely, where to look next. Note that these maps change with successive saccades as posterior beliefs about external states, including the stimulus, become progressively more precise or confident. Note also that salience is depleted in locations that were foveated in the previous saccade because these locations no longer have an epistemic affordance (i.e., the ability to reduce uncertainty or expected free energy). This is a nice illustration of a ubiquitous phenomenon, known as inhibition of return. Oculomotor responses are shown in the third row in terms of the two hidden oculomotor states, corresponding to vertical and horizontal eye movements. The associated portions of the image sampled (at the end of each saccade) are shown in the fourth row. The final two rows show the posterior beliefs in terms of their sufficient statistics and the stimulus categories, respectively. The posterior beliefs are plotted here in terms of conditional expectations and the 90% confidence interval about the true stimulus. The key thing to note here is that the expectation about the true stimulus supervenes over its competing expectations and, as a result, conditional confidence about the stimulus category increases (the confidence intervals shrink to the expectation). This illustrates the nature of evidence accumulation when selecting a hypothesis or percept the best explains sensory data. Within saccade accumulation is evident even during the initial fixation with further stepwise decreases in uncertainty as salient information is sampled at successive saccades. Please see [98] for further details.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 5

  1. Yuan, R., Ma, Y., Yuan, B., and Ping, A. (2010). Bridging Engineering and Physics: Lyapunov Function as Potential Function. arXiv.
  2. Nicolis, G., and Prigogine, I. (1977). Self-Organization in Non-Equilibrium Systems, John Wiley.
  3. Clark, 2013, The many faces of precision, Front. Psychol., 4, 270
  4. Arnold, L. (2003). Random Dynamical Systems (Springer Monographs in Mathematics), Springer.
  5. Velmans, 2008, Reflexive Monism, J. Conscious. Stud., 15, 5

Cited by

See all 189

Metadata

Authors
Friston, K. J., Wiese, W., & Hobson, J. A.
Affiliations
  • Wellcome Centre for Human Neuroimaging (GB)
  • Johannes Gutenberg University Mainz (DE)
  • Harvard University (US)
Corresponding author
  • Friston, Karl J.
  • Wiese, Wanja
  • Hobson, J. Allan
Venue
Entropy
Publisher
MDPI AG
Published
2020-04-30
Volume
22
Issue
5
Pages
516
ISSN
1099-4300 (ISSN-L)
OpenAlex
W3023671316
Cited by
186
References
143
Citations by year
2026: 18 · 2025: 25 · 2024: 31 · 2023: 27 · 2022: 29 · 2021: 40 · 2020: 15
FWCI
26.85
Type
journal article
Language
en
Open access
Locations
  • Entropy (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • UCL Discovery (University College London) (submittedVersion)
Funders
Wellcome Trust (088130/Z/09/Z)
Topics
  • Social Sciences › Arts and Humanities › Philosophy › Philosophy and Theoretical Science
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Embodied and Extended Cognition
  • Social Sciences › Arts and Humanities › History and Philosophy of Science › Philosophy of Biology and Scientific Explanation
Keywords
sentience, consciousness, Cartesian dualism, Markov blanket, information geometry, probability distributions, internal state, origin of consciousness, intentional systems

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.