Verified by 
Environment
The Brain as a Prediction Machine: The Key to Consciousness?
Part 1: Expectations shape many key features of brain and behavior.
Posted January 1, 2022 Reviewed by Vanessa Lancaster
Key points
- Psychology has focused on how the past causes our present behavior. But the future can cause the present too, via prediction and simulation.
- The brain constantly makes predictions. Consciousness enhances the brain's ability to form predictive models of the environment and the self.
- Faulty top-down influence of expectation on perception shapes illusions, weird beliefs, anxiety, depression, psychosomatics, and more.
The future, particularly cognition about the future, has been very much a back-burner issue in psychology for more than a century. The canonical human being, Homo psychologicus, is a prisoner of the past and the present...What happens when the canonical human becomes Homo prospectus, and our ability to think about our futures becomes our defining ability? – Seligman, Railton, Baumeister and Sripada—Homo Prospectus1
Psychology generally, and the scientific study of cognition more specifically, have tended until quite recently to focus more on past-oriented cognition (memory) and present-oriented cognition (perception) and less on future-oriented/prospective cognition (expectation, anticipation).
This has been changing. Increasingly, there is interest in how the brain is oriented to, and perhaps even organized around, forming predictions. And there is interest in developing an evolutionary understanding of why anticipatory cognition is so crucial.
Understanding the brain as a prediction machine may help explain consciousness.
The brain evolved as an adaptation enabling organisms to better perceive and control the environment and their own internal state. Consciousness further enhances this ability by modeling the environment and the self.2 The ability to make predictions, guided by prospective representations—“if-then” possibilities, greatly enhances this modeling ability. A system that can model the environment and itself well can form simulations of the future environment and its adaptations to that environment.
In attempting to resolve the seeming enigma of how the brain produces consciousness, 3 it’s important to first understand "What Actually Is a Thought? And How Is Information Physical?" Thoughts are physical representations or maps. The mind is a kind of map. The brain, and its functional product, the mind, evolved as a map of the body’s relation to its external environment.
More evolved brains, such as human ones, can integrate past sensory experiences to form representations of things that are not presently “out there”—predictive simulations.
Beliefs are a form of predictive modeling.4
When you’re a brain inside a dark skull, guessing what’s presently out there is a form of prediction.
Anil Seth, a neuroscientist at the forefront of the serious science of consciousness,5 explains and develops the concept of the brain as a prediction machine in his book Being You: A New Science of Consciousness.
Seth suggests that even our perceptions of what is presently out there are, in a sense, just predictions or simulations (what he refers to as “controlled hallucinations”). These are formed by a continual process of updating predictions or assumptions (“best guesses”) as to what the sensory data are perceiving.6
As Seth explains in his popular TED talk:
Perception—figuring out what's there—has to be a process of informed guesswork in which the brain combines these sensory signals with its prior expectations or beliefs about the way the world is to form its best guess of what caused those signals. The brain doesn't hear sound or see light. What we perceive is its best guess of what's out there in the world.7
According to this theory, what we perceive is strongly shaped by top-down expectations / predictions, not just by bottom-up sensory input.
How do the brain's thousands of representations/models produce a feeling of unitary consciousness?
In his book, A Thousand Brains,8 the entrepreneur, scientist, engineer, and inventor Jeff Hawkins elaborates on how the brain’s knowledge is stored as numerous models or maps of the world. Research by Hawkins’ group has focused on how the brain has thousands of complementary models of each object it perceives.
He calls this the Thousand Brains Theory of Intelligence. Each model operates as a reference frame, physically stored in a tiny cortical column.9 These complementary reference frames provide “what” and “where” information; acting all together, they provide information about how the different aspects of the object relate to each other.10
Prediction underlies much of the framework for this theory. A cortical column can build powerful predictive models. For example, moving your finger from the bottom of a cup to the top can predict the sensation regardless of how the cup is rotated with respect to the finger sensor.11
So how come we don’t feel like we’re composed of a thousand brains, each with their own independent little model of the world? How come our consciousness feels unified? This is known as the binding problem. Hawkins’ theory proposes that the columns work together through their connections with each other, some of which are long-range connections crisscrossing the entire neocortex.
Through a process akin to “voting,” the different simultaneous models established by different columns encoding perceptions from different reference frames reach a “consensus” best guess (an algorithmic inference) as to what the object is that is being perceived, based on prior learned information (e.g. “the only thing that is consistent with what we’re all perceiving right now, based on the input from all the different senses and from all our different reference frames, is a coffee cup.”).
According to the theory, this system of storing information in reference frames and building models/representations of the world applies not only to physical objects. It scales up at higher levels of complexity and abstraction to concepts like mathematics, politics, and art.
Understanding the central role of expectations in cognition is key to understanding many quirks of human functioning.
The fact that expectations fundamentally shape perceptions and beliefs explains many of the brain’s most successful features, as well as its many problematic bugs.
Seeing is believing. But our perceptions are sometimes wrong. This may lead us to form mistaken (and often, intransigent) beliefs. Equally, believing is seeing—that is, our perceptions are shaped by top-down expectations/assumptions about what we are seeing, hearing, etc. When our prior expectations are mistaken, they powerfully influence us to perceive things in mistaken ways.
This is the basis of optical illusions. Magicians skillfully exploit our perceptual expectations to entertain and amaze us, as do psychics—some of whom are charlatans, others are themselves true believers—victims of self-deception as much as they are responsible for deceiving others.
We, humans, are highly suggestible, which, for many, combines with an incomplete understanding of science and other complex subjects (and an overestimation of that understanding), leading to beliefs in paranormal phenomena, mystical experiences, and other weird beliefs (including conspiracies), confused as to what is real and what is not.
Mistaken expectations play a role in many mental disorders. Anxiety and depression, the most common mental disorders, are characterized by, and some would say partly caused by, faulty expectations of, or predictions about, the future—overestimation of threat or failure.12, 13
Expectations (combined with and shaped by setting/context) also play a major role in the very different ways in which individuals experience psychedelics.14
Furthermore, expectations and unconscious biases in processing bodily sensations play an important role in functional neurological disorders (conversion disorders),15 and in other psychosomatic disorders. And expectations most certainly play a central role in the ubiquitous phenomenon of placebo response.16
Finding purpose in a Godless world
Seligman et al. note17 that the idea that events in nature might be purpose-driven has generally been considered unscientific—teleology has effectively been banished by science. Teleology18 is the idea that design and purpose are inherent in nature. Teleological views of nature predominated in pre-scientific religious worldviews.
Even today, such notions persist in the minds of very many religiously / spiritually-minded people who imagine that a spiritual worldview is compatible with science. A common example is an assumption that one can accept biological evolution and yet believe it to be intended or guided toward some ultimate end. This assumption reveals a deep misunderstanding of evolution.
Both of the major competing twentieth-century theories in psychology—psychoanalytic and behavioral, steered clear of teleology, emphasizing past causation of present behavior and excluding the possibility that any aspect of the future could cause the present.
But, as Seligman et al. conclude:
A conceptual error seems to have animated the lack of interest in the future. Something genuinely suspect—a metaphysical teleology of causation backward in time, of the present by the future—was conflated with something not at all mysterious, namely, the idea that, in a world with minds, behavior can be guided by maps of possible futures as well as the traces of actual pasts. In this way, ‘near future possibilities’ can explain concrete behaviors in the here and now.19
Even though the universe does not have an inherent purpose, living organisms, which evolved spontaneously and unguided on at least this planet and quite probably in many other places in the universe, are purpose-driven. Even the simplest organisms are, by definition, goal-directed. A purposeless universe became infused with purpose, and for humans, imbued with meaning.20
In Part 2 of this two-part post, we explore The Brain as a Prediction Machine: The Key to the Self?
References
1. Martin Seligman, Peter Railton, Roy Baumeister and Chandra Sripada, Homo Prospectus (New York: Oxford University Press, 2016), p.x.
2. Most organisms’ brains lack consciousness or have only very primitive forms of consciousness, by most definitions.
3. Much has been made of what is referred to as the hard problem of consciousness—how subjective, conscious experience can be explained by and reduced to physical processes. As the philosopher David Chalmers, originator of the term “hard problem,” put it: “How can we explain why there is something it is like to entertain a mental image, or to experience an emotion?...Why should physical processing give rise to a rich inner life at all?” David Chalmers, "Facing Up to the Problem of Consciousness," Journal of Consciousness Studies 2, no. 3 (1995), 200-219.
[Click 'More' to view footnotes 4-20].
4. Beliefs are energy-saving shortcuts in modeling and predicting the environment. Beliefs are our brain’s way of making sense of and navigating our complex world. They are mental representations of the ways our brains expect things in our environment to behave, and how things should be related to each other—the patterns our brain expects the world to conform to. Beliefs are templates for efficient learning and are often essential for survival. The brain is an energy-expensive organ, so it had to evolve energy-conserving efficiencies. As a prediction machine, it must take shortcuts for pattern recognition as it processes the vast amounts of information received from the environment by its sense organ outgrowths. Beliefs allow the brain to distill complex information, enabling it to quickly categorize and evaluate information and to jump to conclusions. See “What Actually Is a Belief? And Why Is It So Hard to Change?”
5. The serious science of consciousness needs to be differentiated from the plethora of consciousness pseudoscience that for years has cluttered the field and confused even educated laypeople.
6. The brain employs a process of Bayesian reasoning in doing this—a process of updating prior assumptions, i.e. revising the probabilities, when a new fact is learned or new evidence is observed.
7. He explains further: “If hallucination is a kind of uncontrolled perception, then perception right here and right now is also a kind of hallucination, but a controlled hallucination in which the brain's predictions are being reined in by sensory information from the world. In fact, we're all hallucinating all the time, including right now. It's just that when we agree about our hallucinations, we call that reality.” Anil Seth, “Your brain hallucinates your conscious reality,” TED video transcript, filmed Apr 2017 in Vancouver, BC, https://www.ted.com/talks/anil_seth_your_brain_hallucinates_your_conscious_reality/transcript?language=en.
8. Jeff Hawkins, A Thousand Brains: A New Theory of Intelligence (New York: Basic Books, 2021). See also Jeff Hawkins & Subutai Ahmad | A Thousand Brains: A New Theory of Intelligence | Talks at Google, Aug 7, 2021.
9. Cortical columns span the 2.5 mm thickness of the cerebral cortex and are about 1 mm in area. There are about 150,000 of them, each containing about 100,000 neurons (Hawkins, Talks at Google). According to the Thousand Brains Theory of Intelligence, each cortical column instantiates a reference frame for a given object, with x, y, and z Cartesian spatial coordinates. Each column can also generate behavior—hence they are sensory-motor models. There has been a lack of consensus regarding the exact structure or function of cortical columns, which have generally been regarded as the basic repeating functional units of the neocortex. The Thousand Brains Theory is a proposed solution to the mystery of the exact role of the neocortex's columnar functional organization [Hawkins J, Ahmad S, Cui Y. A Theory of How Columns in the Neocortex Enable Learning the Structure of the World. Front Neural Circuits. 2017;11:81. Published 2017 Oct 25. doi:10.3389/fncir.2017.00081]. Or see here for a short summary of the theory.
10 A single cortical column is of course not solely dedicated to storing information about just one aspect of one object, but rather participates in storing information about aspects of a very large number of different objects.
11. Hawkins J, Lewis M, Klukas M, Purdy S, Ahmad S. A Framework for Intelligence and Cortical Function Based on Grid Cells in the Neocortex. Front Neural Circuits. 2019;12:121. Published 2019 Jan 11. doi:10.3389/fncir.2018.00121
12. Among other tendencies, depressed people tend to imagine fewer positive futures, and they imagine negative future scenarios more quickly and more easily than non-depressed people. They also imagine positive future events less vividly. They tend to overestimate, over-weight, and over-attend to risk, and underestimate their power to influence future events. They have a negative predictive style pertaining to future events, paralleling their negative explanatory style pertaining to present and past events. (Homo Prospectus, Ch, 10)
13. More speculatively, actual hallucinations (“uncontrolled perceptions,” as Seth would call them, as opposed to normal perceptions, which he refers to as “controlled hallucinations”), might perhaps have something to do with the brain’s faulty expectations of its own perceptions—i.e. faulty predictive perceptions (predictive perceptions are analogous to predictive coding in computers). Hallucinations might perhaps be partly due to the brain paying too much attention to its own predictions and too little attention to sensory information contradicting those predictions—a problem in recognizing prediction errors and in updating prior predictions with that information.
See also: Clark A. Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behav Brain Sci. 2013;36(3):181-204. doi:10.1017/S0140525X12000477. And: Jordana Cepelewicz, "To Make Sense of the Present, Brains May Predict the Future," Quanta Magazine, July 2018.
14. Michael Pollan, How to Change Your Mind: What the New Science of Psychedelics Teaches Us about Consciousness, Dying, Addiction, Depression, and Transcendence (New York: Penguin Press, 2018).
15. Functional neurological disorders (FNDs) are disorders presenting with symptoms of altered voluntary motor or sensory function that are incompatible with recognized neurological or medical conditions [American Psychiatric Association DSM-5 Task Force, Diagnostic and Statistical Manual of Mental Disorders: DSM-5, 5th ed. (Washington, DC: American Psychiatric Association, 2013)]. FNDs lack a structural brain cause. They may or may not be associated with psychological stressors as a presumed trigger.
16. Interestingly, the areas of the brain that are dysfunctional in functional neurological disorders are nearly the same areas implicated in placebo effects, most probably due to similar processes involved in both. E.g. the amygdala, a brain structure centrally involved in fear and anxiety, is found to be overactive in functional neurological disorders. Placebo reduces amygdala activity. Burke MJ, Faria V, Cappon D, Pascual-Leone A, Kaptchuk TJ, Santarnecchi E. Leveraging the Shared Neurobiology of Placebo Effects and Functional Neurological Disorder: A Call for Research. J Neuropsychiatry Clin Neurosci. 2020;32(1):101-104. doi:10.1176/appi.neuropsych.19030077.
17. Seligman ME, Railton P, Baumeister RF, Sripada C. Navigating Into the Future or Driven by the Past. Perspect Psychol Sci. 2013;8(2):119-141. doi:10.1177/1745691612474317.
18. from the Greek word “telos,” for “end” or “purpose.”
19. Homo Prospectus, p.352.
20. For a book-length explanation of how a purposeless universe became infused with purpose, and for us humans, imbued with meaning, see Ralph Lewis, Finding Purpose in a Godless World: Why We Care Even If The Universe Doesn’t (Amherst, NY: Prometheus Books, 2018).
