Skip to main content
Log in

Are Information or Data Patterns Correlated with Consciousness?

  • Published:
Topoi Aims and scope Submit manuscript

Abstract

Scientific research on consciousness is attempting to gather data about the relationship between consciousness and the physical world. The basic procedure is to measure consciousness through first-person reports, measure the physical world and look for correlations between these sets of measurements. While this work has focused on neural correlates of consciousness, it has also been proposed that information states in the brain might be linked to consciousness. This paper uses Floridi’s distinction between dedomena, data and information to state this claim more precisely and suggests that the best starting point for this work is the potential correlation between data patterns and consciousness. Floridi’s method of levels of abstraction is used to understand how data can be measured at different levels of abstraction in the brain and the paper examines a number of problems with this work, which could make it difficult to prove that data patterns are correlated with consciousness.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6

Similar content being viewed by others

Notes

  1. Some of these potential correlates are described in a review paper by Tononi and Koch (2008).

  2. Some people have expressed scepticism about the idea that there could be correlates of consciousness or that it is possible to identify them in the brain (Molyneux 2010; Noë and Thompson 2004). Some of the methodological issues with the scientific study of consciousness are discussed by Gamez in ‘The Measurement of Consciousness’ (currently under review).

  3. Chalmers (2000) distinguishes the total neural basis from the core neural basis: “A total NCC builds in everything and thus automatically suffices for the corresponding conscious states. A core NCC, on the other hand, contains only the ‘core’ processes that correlate with consciousness. The rest of the total NCC will be relegated to some sort of background conditions required for the correct functioning of the core.” (Chalmers 2000, p. 26). Block (2007) makes a similar distinction.

  4. While there has been a substantial amount of speculation about the connection between computations in the brain and consciousness, few attempts have been made to specify what it means for a brain to implement a computation. This is necessary if one wants to carry out experiments on the computational correlates of consciousness. Chalmers’ (2011) combinatorial state automata account is arguably the best theory of implementation that has been put forward. However, it has a number of problems, which I have discussed elsewhere in “Are there Functional Correlates of Consciousness?”, currently under review. A draft copy is available on request.

  5. The terms stateless Φ and state-based Φ were introduced in a previous paper to distinguish Tononi’s two measures of information integration (Gamez and Aleksander 2011).

  6. While Tononi and Sporns’ (2003) algorithm identifies a single main complex, Balduzzi and Tononi’s (2008) algorithm can potentially identify more than one main complex in a system. In this paper I will speak as if information integration algorithms only identify a single main complex.

  7. Functional connectivity (a deviation from statistical independence between A and B) is typically contrasted with anatomical connectivity (a physical link between A and B) and from effective connectivity (a causal link from A to B).

  8. This lack of relationship between some levels of abstraction in the brain is likely to prevent Floridi’s (2008) concept of a gradient of abstraction from being applied to the problems discussed in this paper.

  9. Since data patterns are themselves data sets, data sets and data patterns could be combined into a hierarchy of levels of abstraction—what Floridi (2008) calls a gradient of abstraction. However, in experiments on the correlates of consciousness there is an important practical distinction between measurements of the system (using EEG, electrodes, fMRI, etc.) and the ways in which these measurements are processed by algorithms to identify patterns that might be correlated with consciousness. To preserve this distinction and minimise confusion, I will use ‘data sets’ to describe the different ways in which scientists extract measurements of the physical world at a level of abstraction, and talk about the possibility that patterns in these data sets might be correlated with consciousness (H2).

  10. The mathematics of topology could be used to provide a more precise definition of spatial pattern matching, but this is beyond the scope of this paper.

  11. A different way of addressing this problem would be to assume that a sleeping or anaesthetized brain is actually conscious, but unable to report or remember its consciousness. However, this would make contrastive experiments on the correlates of consciousness impossible because there will not be states of the system in which consciousness is known to be absent. I have discussed these issues elsewhere in ‘The Measurement of Consciousness’, currently under review. A draft is available on request.

  12. Tononi (2010) suggests this approach.

  13. It has been suggested that we could demonstrate substrate independence by replacing part or all of the brain with a functionally equivalent silicon chip (Chalmers 1995). This would hold the data patterns constant while the underlying substrate was changed. If the system continued to be conscious, then we could claim that consciousness was correlated with data patterns, and not with a pattern in a particular physical substrate. The problem with this experiment is that the subject would behave in exactly the same way before and after the implantation of the chip. The only thing that could change during the experiment is our interpretation of their reports about consciousness. Either we would continue to believe that their reports were measurements of consciousness or we would believe that the substitution had removed the subject’s capacity for consciousness and that the reports were zombie behaviour. Since the outcome of the experiment depends on the experimenter’s prior assumptions about the brain, it would not be a valid demonstration that there are substrate-independent data correlates of consciousness. A more detailed discussion of this point is given by Gamez (2012a).

  14. While we are unlikely to conclusively prove that a data pattern, d 1 , is correlated with consciousness in a way that is consistent with R1–R3, we might still be able to show that d 1 is present at many levels of abstraction when the brain is conscious and not present at many levels of abstraction when the brain is unconscious. In this case it could be claimed that d 1 is a likely correlate of consciousness or the ‘best game in town’ for a correlate of consciousness.

    If experiments have shown that d 1 has some substrate independence (R3), then it would be reasonable to claim that there is empirical support for a correlation between a data pattern and consciousness, and we could treat all systems in which d 1 can be found as potentially conscious. However, if there is no evidence for the substrate independence of d 1 , then the experiments could also be interpreted to show that a pattern in a particular physical structure in the brain is correlated with consciousness - for example, a neural pattern or pattern in electromagnetic waves. In this case we should suspend judgement about whether d 1 forms a CC set by itself until evidence for substrate independence has been found.

  15. Normalization can introduce instabilities in the Φ value of the subset—see Barrett and Seth (2011).

References

  • Ahrens MB, Orger MB, Robson DN et al (2013) Whole-brain functional imaging at cellular resolution using light-sheet microscopy. Nat Methods 10(5):413–420

    Article  Google Scholar 

  • Aru J, Bachmann T, Singer W et al (2012) Distilling the neural correlates of consciousness. Neurosci Biobehav Rev 36(2):737–746

    Article  Google Scholar 

  • Balduzzi D, Tononi G (2008) Integrated information in discrete dynamical systems: motivation and theoretical framework. PLoS Comput Biol 4(6):e1000091

    Article  Google Scholar 

  • Balduzzi D, Tononi G (2009) Qualia: the geometry of integrated information. PLoS Comput Biol 5(8):e1000462

    Article  Google Scholar 

  • Barrett AB, Seth AK (2011) Practical measures of integrated information for time-series data. PLoS Comput Biol 7(1):e1001052

    Article  Google Scholar 

  • Block N (2007) Consciousness, accessibility, and the mesh between psychology and neuroscience. Behav Brain Sci 30(5–6):481–499

    Google Scholar 

  • Casali AG, Gosseries O, Rosanova M et al (2013) A theoretically based index of consciousness independent of sensory processing and behavior. Sci Transl Med 5(198):198ra105

    Google Scholar 

  • Chalmers D (1995) Absent qualia, fading qualia, dancing qualia. In: Metzinger T (ed) Conscious experience. Imprint Academic, Thorverton

    Google Scholar 

  • Chalmers D (2000) What is a neural correlate of consciousness? In: Metzinger T (ed) Neural correlates of consciousness. MIT Press, Massachusetts

    Google Scholar 

  • Chalmers D (2010) The singularity: a philosophical analysis. J Conscious Stud 17:7–65

    Google Scholar 

  • Chalmers D (2011) A computational foundation for the study of cognition. J Cognit Sci 12(4):323–357

    Google Scholar 

  • de Graaf TA, Hsieh PJ, Sack AT (2012) The ‘correlates’ in neural correlates of consciousness. Neurosci Biobehav Rev 36(1):191–197

    Article  Google Scholar 

  • Dehaene S, Naccache L, Cohen L et al (2001) Cerebral mechanisms of word masking and unconscious repetition priming. Nat Neurosci 4(7):752–758

    Article  Google Scholar 

  • Ferrarelli F, Massimini M, Sarasso S et al (2010) Breakdown in cortical effective connectivity during midazolam-induced loss of consciousness. Proc Natl Acad Sci USA 107(6):2681–2686

    Article  Google Scholar 

  • Floridi L (2008) The method of levels of abstraction. Mind Mach 18(3):303–329

    Article  Google Scholar 

  • Floridi L (2009) Philosophical conceptions of information. Lect Notes Comput Sci 5363:13–53

    Article  Google Scholar 

  • Gamez D (2011) Information and consciousness. Etica & Politica/Ethics & Politics XIII(2):215–234

    Google Scholar 

  • Gamez D (2012a) Empirically grounded claims about consciousness in computers. Int J Mach Conscious 4(2):421–438

    Article  Google Scholar 

  • Gamez D (2012b) From Baconian to Popperian neuroscience. Neural Syst Circuits 2(1):2

    Article  Google Scholar 

  • Gamez D, Aleksander I (2011) Accuracy and performance of the state-based Φ and liveliness measures of information integration. Conscious Cogn 20(4):1403–1424

    Article  Google Scholar 

  • Granger C (1969) Investigating causal relations by econometric models and cross-spectral methods. Econometrica 37:424–438

    Article  Google Scholar 

  • Hadley MW, McGranaghan MF, Willey A et al (2012) A new measure based on degree distribution that links information theory and network graph analysis. Neural Syst Circuits 2(1):7

    Article  Google Scholar 

  • Hubel DH, Wiesel TN (1959) Receptive fields of single neurones in the cat’s striate cortex. J Physiol 148:574–591

    Article  Google Scholar 

  • Koch C (2004) The quest for consciousness: a neurobiological approach. Roberts & Company, Englewood

    Google Scholar 

  • Kurzweil R (1999) The age of spiritual machines. Viking, New York

    Google Scholar 

  • Lee U, Mashour GA, Kim S et al (2009) Propofol induction reduces the capacity for neural information integration: implications for the mechanism of consciousness and general anesthesia. Conscious Cogn 18(1):56–64

    Article  Google Scholar 

  • Logothetis NK (1998) Single units and conscious vision. Philos Trans R Soc Lond B Biol Sci 353(1377):1801–1818

    Article  Google Scholar 

  • Massimini M, Boly M, Casali A et al (2009) A perturbational approach for evaluating the brain’s capacity for consciousness. Prog Brain Res 177:201–214

    Article  Google Scholar 

  • Molyneux B (2010) Why the neural correlates of consciousness cannot be found. J Conscious Stud 17(9–10):168–188

    Google Scholar 

  • Noë A, Thompson E (2004) Are there neural correlates of consciousness? J Conscious Stud 11(1):3–28

    Google Scholar 

  • Popper KR (2002) The logic of scientific discovery. Routledge, London

    Google Scholar 

  • Quiroga RQ, Reddy L, Kreiman G et al (2005) Invariant visual representation by single neurons in the human brain. Nature 435(7045):1102–1107

    Article  Google Scholar 

  • Schreiber T (2000) Measuring information transfer. Phys Rev Lett 85(2):461–464

    Article  Google Scholar 

  • Seth AK, Izhikevich E, Reeke GN et al (2006) Theories and measures of consciousness: an extended framework. Proc Natl Acad Sci USA 103(28):10799–10804

    Article  Google Scholar 

  • Shanahan M, Wildie M (2012) Knotty-centrality: finding the connective core of a complex network. PLoS ONE 7(5):e36579

    Article  Google Scholar 

  • Shannon CE (1948) A mathematical theory of communication. Bell Syst Techn J 27(379–423):623–656

    Article  Google Scholar 

  • Teasdale G, Jennett B (1974) Assessment of coma and impaired consciousness. A practical scale. Lancet 2(7872):81–84

    Article  Google Scholar 

  • Tononi G (2004) An information integration theory of consciousness. BMC Neurosci 5:42

    Article  Google Scholar 

  • Tononi G (2008) Consciousness as integrated information: a provisional manifesto. Biol Bull 215(3):216–242

    Article  Google Scholar 

  • Tononi G (2010) Information integration: its relevance to brain function and consciousness. Arch Ital Biol 148(3):299–322

    Google Scholar 

  • Tononi G, Koch C (2008) The neural correlates of consciousness: an update. Ann N Y Acad Sci 1124:239–261

    Article  Google Scholar 

  • Tononi G, Sporns O (2003) Measuring information integration. BMC Neurosci 4:31

    Article  Google Scholar 

  • Tononi G, Sporns O, Edelman GM (1994) A measure for brain complexity: relating functional segregation and integration in the nervous system. Proc Natl Acad Sci USA 91(11):5033–5037

    Article  Google Scholar 

  • Wilkes KV (1988) yìshì, duh, um, and consciousness. In: Marcel AJ, Bisiach E (eds) Consciousness in Contemporary Science. Clarendon Press, Oxford

    Google Scholar 

Download references

Acknowledgments

I would like to thank Barry Cooper and the John Templeton Foundation for supporting this work (Project ID 15619: ‘Mind, Mechanism and Mathematics: Turing Centenary Research Project’). I would also like to thank Anil Seth and the Sackler Centre for Consciousness Science at the University of Sussex for hosting me as a Research Fellow during this project. I am grateful to the referees for their helpful comments, which have considerably improved this paper.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to David Gamez.

Appendix 1: Balduzzi and Tononi’s ‘Information Integration’ Algorithm

Appendix 1: Balduzzi and Tononi’s ‘Information Integration’ Algorithm

Balduzzi and Tononi’s (2008) ‘information integration’ algorithm is applied to a network of elements that are in a particular state. These elements are particular parts or aspects of a system that are linked to observables at a level of abstraction. The states of the observables are the values in the typed variables that change as the measured aspect of the system changes. Different levels of abstraction lead to different sets of observables that correspond to different elements in the system. Although this is actually a data integration algorithm (see Sect. 3.3), the summary in this appendix uses the same terminology as the original paper.

Balduzzi and Tononi’s (2008) algorithm uses relative entropy to measure the effective information that is generated by a subset of the elements when they enter a particular state. The relative entropy, H[p||q], between probability distributions p and q is given by Eq. 1:

$$ H[p||q] = \sum\limits_{i} {p_{i} \log_{2} \frac{{p_{i} }}{{q_{i} }}} . $$
(1)

When a set of elements are in a particular state, x 1 , at time t 1 there is a certain probability that each possible state at t 0 led to the current state at t 1 . The set of these probabilities is called the a posteriori repertoire or p(X 0  → x 1 ). The entering of the elements into state x 1 generates information because it defines the a posteriori repertoire of probabilities of the possible states that could have existed at t 0 and led to the current state. However, if the elements are in a state of maximum entropy and their states are entirely random, then each state x 1 could have been caused by any other state, and the fact that you are in x 1 tells you nothing about the previous state. In this case, the probability that each possible previous state of the elements caused the current state is the same, and this set of probabilities is known as the a priori repertoire, or p max(X 0 ). According to Balduzzi and Tononi (2008), the amount of information generated by a particular state can be measured using the relative entropy between the a posteriori repertoire associated with x 1 and the a priori repertoire, as expressed in Eq. 2:

$$ ei(X_{0} \to x_{1} ) = H[p(X_{0} \to x_{1} )||p^{\hbox{max} } (X_{0} )] , $$
(2)

where ei(X 0  → x 1 ) is the effective information generated by the state x 1 , p(X 0  → x 1 ) is the a posteriori repertoire and p max(X 0 ) is the a priori repertoire.

Equation 2 gives the effective information that is generated when some or all of the elements enter a particular state, but it does not tell us whether this information was generated by causal interactions among the elements, or whether it is the sum of the information generated by the elements acting independently. To answer this question Balduzzi and Tononi (2008) consider partitions of the system and calculate the relative entropy between the a posteriori repertoires of the parts considered independently and the a posteriori repertoire of the whole subset, as expressed in Eq. 3:

$$ ei(X_{0} \to x_{1} /P) = H\left[ {p(X_{0} \to x_{1} )\left\| {\prod\limits_{{M^{k} \in P^{{}} }} {p(M_{0}^{k} \to \mu_{1}^{k} )} } \right.} \right] , $$
(3)

where ei(X 0  → x 1 /P) is the effective information of a particular partition, P, of the system into two or more parts, M k is a part of the system, and μ k is a state of M k. To calculate ei(X 0  → x 1 /P), each part is considered as a system in its own right and the inputs from the other parts are treated as noise.

The minimum information partition is the subset or division of the subset across which the least information is integrated, and it is identified by comparing the normalized effective information values for each possible partition as well as the normalized effective information for the subset as a whole (known as the total partition, whose effective information is calculated using Eq. 2). Normalization is needed during this comparison process because the effective information across a partition between a single element and a number of elements is typically less than an equal bipartition, and the effective information across many partitions is typically higher than the effective information across few partitions.Footnote 15 The un-normalized value of effective information for the minimum information partition is the Φ value of the subset, which is calculated for every possible subset of the system.

Balduzzi and Tononi (2008) define a complex as a subset that is not included in another subset with higher Φ. According to Balduzzi and Tononi, complexes are regions of the system where elements integrate the most information, and the Φ value of each complex corresponds to the amount of information that is integrated. The main complexes of the system are the complexes whose subsets have strictly lower Φ, and Tononi (2004, 2008) claims that main complexes are the conscious parts of the system. A correlation-based interpretation of information integration would interpret the main complexes as the parts of the system that are predicted to be correlated with consciousness.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Gamez, D. Are Information or Data Patterns Correlated with Consciousness?. Topoi 35, 225–239 (2016). https://doi.org/10.1007/s11245-014-9246-7

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11245-014-9246-7

Keywords

Navigation