Abstract
Predictive processing is an increasingly popular approach to cognition, perception and action. It says that the brain is essentially a hierarchical prediction machine. It is typically construed in a representationalist and inferentialist fashion so that the brain makes contentful inferences on the basis of representational models. In this paper, I argue that the predictive processing framework is inconsistent with this epistemic position. In particular, I argue that the combination of hierarchical modeling, contentful inferentialism and representationalism entail an internal inconsistency. Specifically, for a particular set of states, there will be both a representation requirement and not. Yet a system cannot both be required to represent a certain set of states and not be required to represent those states. Due to this contradiction, I propose to reject the standard view. I suggest that predictive processing is best interpreted in terms of reliable covariation instead, entailing an instrumentalist approach to the statistical machinery.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
The term predictive processing was first introduced in Clark (2013) to distinguish the action-oriented approach defended there from the more passive perceptual approach linked to the term predictive coding.
The formulation may seem redundant. It may seem that if a system is inferential, it is necessarily also representational. We will see below that, in the PP and FEP literature, this is not always the case due to a particular technical notion of inference.
Since some of the most influential works in PP published in 2013 (Friston 2013; Hohwy 2013; Clark 2013), the literature has expanded rapidly. I will discuss Hohwy’s seclusionist approach and touch on a more recent approach proposed by Kirchhoff, Ramstead and colleagues (see references in text) that rejects neurocentrism, seclusionism, and arguably representationalism (more on this in Section 4) as well as a radical enactive proposal by Gallagher and Allen (2018) that suggests rebranding the approach to predictive engagement.
It is important to point out that this is a vital point of disagreement between later FEP theorists and Hohwy. Though Hohwy concedes that active inference is vital to maintain the organism in the long run, and perceptual inference can help alleviate prediction error short term, he regards neither as superior to the other (Hohwy 2016, p. 280; Hohwy 2017). Later FEP theorists will argue that perceptual inference is only a step in the process of active inference, which maintains the organism in healthy bounds for as long as possible (e.g. Kirchhoff and Kiverstein 2019 and Bruineberg et al. 2016).
This opposed to views of direct perception, in which the body is supposed to be in direct contact with the world with no space for an evil demon to intervene. In Hohwy (2017) (endearingly titled How to Entrain your Evil Demon) he explicates the ‘wriggling space’ of the evil demon more precisely. Roughly, by emphasizing the importance of active inference in sampling the world selectively, he minimizes the role the evil demon could play. This does not however affect the relevant principled commitments.
This showcases the appeal of PP. A single computational pattern is thought to be found at all levels of self-organization, from single cells, via the neural architecture to full-blown organisms, and other times even niche construction, sociality and human cultural practices (see Hesp et al. 2019 for an overview). It is important to note that Hohwy typically prefers to remain neurocentric (Hohwy 2016).
Relevant to the concept of a Markov blanket are many statistical equations. As this paper aims to make a purely conceptual point, I will not include a mathematical description of the formalism. See Friston (2010, 2013) and Friston and Stephan (2007) for a mathematically informed introduction into the formalism and its use in FEP.
Whether or not to cash this out in terms of structural representations or not is currently a hot debate (Kirchhoff and Robertson 2018; Bruineberg et al. 2016; Gładziejewski 2016; Williams 2018). I shall argue that, to avoid the objections raised in this paper, we will have to rid ourselves of representations.
This picture is close to what is being proposed in the FEP literature, where the world is described as consisting of Markov blankets of Markov blankets (Ramstead et al. 2016, 2018; Kirchhoff et al. 2018). We will delve more into this view below, and how it relates to Hohwy’s view. We will see that a few important theoretical posits on Hohwy’s side make the difference.
The original context of this quote is Hohwy’s discussion of the possibility of extended cognition in PP, in which the proliferation of agents is brought up as one of a few reasons for the notion of extended cognition to be unappealing under PP. Further on he remarks that this objection also holds for a neurocentric approach, after which he discusses some possible solutions I discuss in text below.
I would argue that the system cannot and need not represent the target, but full exposition of the argumentation is outside of the scope of this paper.
Cartesian skepticism and seclusionism most likely stand or fall together with inferentialism and representationalism, but full exposition of these ideas is outside of the scope of this paper.
There may be other good reasons to prefer representationalism or Hohwy’s take on inferentialism (I would argue there aren’t, but that is outside of the scope of this paper). Here I restrict myself to what follows immediately out of the Markov blanket formalism and show that the formalism itself does not entail such an interpretation of the target system.
Technically, one may choose to invoke representations at higher hierarchical levels, yet this is not necessary. Moreover, the inconsistency may well reappear as long as there are still multiple hierarchical levels at which representation is required.
De Oliveira (2018) has argued for a non-representational approach to models in science. This notion of modeling explains the origin and use of modeling in terms of its surrogacy for the target system and our scientific practices of modeling.
References
Anderson, M. (2017). Of Bayes and bullets: An embodied, situated, targeting-based account of predictive processing. In T. Metzinger & W. Wiese (Eds.), Philosophy and predictive processing. Frankfurt am Main: MIND Group.
Bruineberg, J., & Rietveld, E. (2014). Self-organization, free energy minimization and an optimal grip on a field of affordances. Frontiers in Human Neuroscience, 8, 599.
Bruineberg, J., Kiverstein, J., & Rietveld, E. (2016). The anticipating brain is not a scientist: The free-energy principle from an ecological-enactive perspective. Synthese, 195, 2417.
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36, 181–253.
Clark, A. (2016). Surfing uncertainty: Prediction, action, and the embodied mind. Oxford: Oxford University Press.
Constant, A., Ramstead, M. J., Veissière, S. P., Campbell, J. O., & Friston, K. (2018). A variational approach to niche construction. Journal of the Royal Society Interface, 15, 141.
De Oliveira (2018) Representationalism is a dead end. Synthese, 1–27.
Dennett, D. C. (2013). Intuition pumps and other tools for thinking. New York: W.W. Norton & Company.
Engel, A. K., Fries, P., & Singer, W. (2001). Dynamic predictions: Oscillations and synchrony in top-down processing. Nature Reviews Neuroscience, 2(10), 704–716.
Fabry, R. (2017). Transcending the evidentiary boundary: Prediction error minimization, embodied interaction, and explanatory pluralism. Philosophical Psychology, 30(4), 395–414.
Frigg, R., & Hartmann, S. (2018). Models in science. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Summer 2018 Edition), URL = https://plato.stanford.edu/archives/sum2018/entries/models-science/
Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138.
Friston, K. (2012). A free energy principle for biological systems. Entropy, 14, 2100–2121.
Friston, K. (2013). Life as we know it. Journal of the Royal Society, Interface, 10, 20130475.
Friston, K., & Siebel, S. (2009). Predictive coding under the free-energy principle. Philosophical Transactions of the Royal Society B, 364, 1211–1221.
Friston, K., Daunizeau, J., Kilner, J., & Kiebel, S. (2010). Action and behavior: A free–energy formulation. Biological Cybernetics, 102(3), 227–260.
Friston, K., Samothrakis, S., & Montague, R. (2012). Active inference and agency: Optimal control without cost functions. Biological Cybernetics, 106(8), 523–541.
Friston, K. and Stephan, K. (2007). Free energy and the brain. Synthese 159(3): 417–458.
Gallagher, S. (2017). Enactivist interventions: Rethinking the mind. Oxford: Oxford University Press.
Gallagher, S., & Allen, M. (2018). Active inference, enactivism and the hermeneutics of social cognition. Synthese, 195(6), 2627–2648.
Gładziejewski, P. (2016). Predictive coding and representationalism. Synthese, 193(2), 559–582.
Hesp, C., Ramstead, M. J., Constant, A., Badcock, P. B., Kirchhoff, M. D., & Friston, K. J. (2019). A multi-scale view of the emergent complexity of life: A free-energy proposal. In M. E. Price et al. (Eds.), Evolution, development, and complexity: Multiscale models in complex adaptive systems. Cham: Springer.
Hohwy, J. (2013). The predictive mind. Oxford: Oxford University Press.
Hohwy, J. (2016). The self-evidencing brain. Noûs, 50(2), 259–285.
Hohwy, J. (2017). How to entrain your evil demon. In T. Metzinger & W. Wiese (Eds.), Philosophy and predictive processing. Frankfurt am Main: MIND Group.
Hutto, D. D., & Myin, E. (2013). Radicalizing enactivism: Basic minds without content. Cambridge: MIT Press.
Hutto, D. D., & Myin, E. (2017). Evolving enactivism: Basic minds meet content. Cambridge: MIT Press.
Kirchhoff. (2018a). Autopoiesis, free energy and the life mind continuity thesis. Synthese, 195, 2519–2540.
Kirchhoff. (2018b). Predictive processing, perceiving and imagining: Is to perceive to imagine, or something close to it. Philosophical Studies, 175, 751–767.
Kirchhoff. (2018c). The body in action: Predictive processing and the embodiment thesis. In A. Newen, L. De Bruin, & S. Gallagher (Eds.), Oxford handbook of 4E cognition: Embodied, extended and enactive. Oxford: Oxford University Press.
Kirchhoff, M., & Robertson, I. (2018). Enactivism and predictive processing: A non-representational view. Philosophical Explorations, 21(2), 264–281.
Kirchhoff, M., Parr, T., Palacios, E., Friston, K., & Kiverstein, J. (2018). The Markov blankets of life: Autonomy, active inference and the free energy principle. Journal of the Royal Society, Interface, 15, 20170792.
Kirchhoff, M. and Kiverstein, J. (2019) Extended Consciousness and Predictive Processing: A Third Wave View. Routledge.
Marr, D. (1982). Vision: a computational investigation into the human representation and processing of visual information. Freeman.
Ramstead, M., Veissière, S., & Kirmayer, L. (2016). Cultural affordances: scaffolding local worlds through shared intentionality and regimes of attention. Frontiers in Psychology, 7(1090).
Ramstead, M., Badcock, P., & Friston, K. (2018). Answering Schrödinger’s question: a free-energy formulation. Physics of Life Reviews, 24, 1–16.
Ramstead, M., Kirchhoff, M., & Friston, K. (2019). A tale of two densities: Active inference is enactive inference. Adaptive Behavior, 1–15.
Rao, R. P., & Ballard, D. H. (1999). Predictive coding in the visual cortex: A functional interpretation of some extra-classical receptive-field effects. Nature Neuroscience, 2(1), 79–87.
Seth, A. (2013). Interoceptive inference, emotion and the embodied self. Trends in Cognitive Sciences, 17(11), 565–573.
Travis, C. (2004). The silence of the senses. Mind, 113(449), 57–94.
Wiese, W., & Metzinger, T. (2017). Vanilla PP for philosophers: A primer on predictive processing. In T. Metzinger & W. Wiese (Eds.), Philosophy and predictive processing. Frankfurt am Main: MIND Group.
Williams, D. (2018). Predictive processing and the representation wars. Minds and Machines, 28, 141.
Acknowledgements
This paper was funded by the FWO (Grant No. 1124818 N). I thank two anonymous reviewers, Erik Myin, Farid Zahnoun and Victor Loughlin for helpful commentary on earlier versions of this paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
van Es, T. Minimizing prediction errors in predictive processing: from inconsistency to non-representationalism. Phenom Cogn Sci 19, 997–1017 (2020). https://doi.org/10.1007/s11097-019-09649-y
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11097-019-09649-y