Book review: Graziano's Rethinking Consciousness

Michael Graziano's Rethinking Consciousness, A Scientific Theory of Subjective Experience [0] is a popular science book where he elaborates in a non-technical way about his attention schema theory and concludes the book discussing implications for artificial intelligence and uploading minds.  During the journey, he explains his materialistic view of what consciousness is.  His definition is that consciousness  is the ability to have a subjective experience.  He explains that through neuroscience, consciousness, or one's subjective experience, arises from the attention schema which constructs a rough simple model of the world which determines what stimuli one's attention should focus on.  This interplay between stimuli, the underlying neural network, and the attention schema creates the feeling that one has a mind or subjective experience.  Consciousness is not an illusion nor is it something non-physical such as a soul or essence, but rather, it emerges from the ingredients similarly to temperature and pressure emerge from the average speed of atoms in a gas in a container.

This first part of explaining the foundation necessary for the attention schema theory and how that theory actually works is somewhat confusing and left me with a feeling that somehow it was all begging the question...the conclusion was somehow assumed in the premises of the argument and it ended up in a big circle.  But the next part speculating on the application to artificial (and given enough technical progress, biological) intelligence is worth the slog.  He posits that there are 4 essential elements to a being that can have consciousness or a subjective experience: (1) be able to pay attention to something; (2) an attention schema that can shift the attention between different things; (3) a wide range of content (memories, data, etc.); and (4) the ability to search the content to find relevant facts on which the attention schema can make decisions to direct the attention.  Following this explanation is a survey of the state of technology (in 2019) to be able to instantiate the necessary pieces to construct an artificial intelligence per this blueprint.  In the discussion, emotional states and body awareness are mentioned as two possible additional necessary elements to arrive at consciousness as we know it.  Graziano speculates these could be added virtually for artificial intelligences, but need more consideration in the biological scenario.

This deconstruction of the necessary elements of consciousness, Graziano suggests, could provide a means to test for it.  Certainly, a computer program that was written to be an artificial intelligence could be examined to see if it in fact fulfilled the requirements.  Biological wetware is harder to examine, but morphological differences between brains would show whether the structures where we currently believe the processes take place are present.  As scanning resolution increases (cue the alien abduction visual), scanning and probing could allow verification that these essential elements are present.  Based on morphology, he asserts that only the higher mammals have consciousness.  One of the consequences of what Graziano proposes is that consciousness is algorithmic.  So, on the question of Searle's Chinese Room paradox, it seems Graziano would argue that the room (not the person within), is conscious. [1]  Thus any Universal Turing Machine can become conscious given the right algorithm and enough data.

The last portion of the book focuses on the possibility of uploading one's mind, whether the Matrix is possible, and the ethical challenges with duplicate "selfs", if such are ever constructed.  While interesting, given how far technology has to go, it remains firmly in the realm of science fiction.

My motivation for reading this book came from discussions on the topic of free will.  Graziano offers no opinion overtly.  The materialistic approach of the explanation suggests he might fall into the determinist camp although he does offer that consciousness is not necessary for decision making [2] and usually, it is argued that consciousness is necessary for free will.  And so, I will end it here and I have Rescher's Free Will [3] on my night stand, borrowed through interlibrary loans from the University of Idaho.  I am in the middle of chapter 1 and already have a lot of reactions to put into words.  (stated that way as a preview of what's to come!)

[0] Graziano, Michael S A. Rethinking Consciousness: A Scientific Theory of Subjective Experience. United States, W. W. Norton, 2019.

[1] https://plato.stanford.edu/entries/chinese-room/

[2] Graziano, p.6

[3] Rescher, Nicholas. Free Will: A Philosophical Reappraisal. United Kingdom, Transaction Publishers, 2009.

Book Review: Free Will by Sam Harris

 Sam Harris' Free Will [0] is a thin volume (one of the jacket blurb mentions 13,000 words) that is easily digested in an afternoon.  He sets out to show that two assumptions are false: "(1) that each of us could have behaved differently than we did in the past, and (2) that we are the conscious source of most of our thoughts and actions in the present."  Both of these are approached from a neuroscience perspective.  Neuroscience has made progress in mapping conscious and unconscious mental states and processes to physical brain states and changes. [1] Some of this comes from analysis of people with brain injuries. [2]

Examining through introspection, it seems that the thoughts that occur are self-generated.  The way thoughts appear to ourselves is highly suggestive that we have free will.  However, since much of what happens in the brain is not accessible to introspection and we know that some of the heuristics used by the brain can be easily manipulated [3], the introspective conclusion is suspect at best.  For example, having sat on the beach observing carefully for ideas for next year's sandcastle building competition, one would also notice the movement of the sun over the course of the day and conclude that the sun moves around the earth.  But of course this is an illusion.  Physiologically, Harris argues, the conscious thoughts and decisions come from the unconscious which has been ultimately influenced by externalities and therefore no thought or decision can truly be created ex nihilo.

The question that arises here is whether the existence of prior influences are sufficient that thoughts and decisions are pre-determined or is there room for a person to have bechaved differently in a given situation. A die thrown with a certain velocity and spin onto a specific surface will stop with the same number on top under Newtonian mechanics.  That we may not be able to predict it because we do not know the specific values of the velocity or spin does not negate the fact that repeating the throw exactly will not alter the result.  The sources and effects of the pre-determining influences on our thoughts are vast and diverse so as to make it currently impossible to determine.  (Although, one might consider "psychohistory" envisioned in Asimov's Foundation series as a basis for a methodology. [4])

The definitive experiment would be to run time forward and backward over some point in time to see if indeed one behaved differently.  Again, under Newtonian mechanics as well as Einstein's relativistic mechanics, outcomes are pre-determined and therefore would never vary although they may be difficult to compute.  Quantum behaviour, such as where an electron will hit a screen after passing through a double slit, is not pre-determined although there are regions where the probability is zero.  For the regions with non-zero probability, there is no calculation made or process used that can narrow the result a given trial any further than the computed probability density.  For many, this has given hope that if the physiological mechanics occur at scales sufficiently small that quantum effects can result in different outcomes for the same set of inputs, then a physiological basis for free will is established.  Unfortunately this may be a pyrrhic victory in that one does not consciously think and decide upon a course of action, it is merely chosen for us through the arbitrary collapse of a wave function somewhere amongst the myriad of neurons in one's brain.

Another argument that gets raised is the emergent complex behaviour of simple systems such as, for example, cellular automata [5]: the "self" that thinks thoughts and decides decisions is something arising out of the complexity of the neural network in the brain and is distinct from the underlying substructure.  But this emergent behaviour is not something that exists of itself and is merely pareidolia no more autonomous than the patterns of dots occurring in the cellular automata simulation.  Of course there is a difference between simple cellular automata and a human brain in that the brain does exhibit the ability to learn and change behaviour based on this learning.  But learning should not be mistaken for free will, it is an improvement in the ability to predict outcomes and thus is another input into the calculation of what is next thought or decided.

Harris dismisses the compatibilist view as simply a post hoc justification of what has occurred which does not substantiate that an alternate history could have happened.  When it is asserted that someone or something, other than the self, has free will, this is based on observation that it behaves in ways that are self-generated, goal-seeking, and non-uniform in response the same stimuli.  However, rule-basead "artificial intelligences" such as computer opponents in various games also display these traits as do most animals.  Assertions about free will are about interior states not amenable observation and where self-reporting is suspect as explained above.  As a result, when discussing the past, at best one can provide an enumeration of roads not taken, and when discussing the future, it devolves into an analysis of constraints on free action.  Far from making the case for free will, it is not even wrong.

Thus Harris succeeds in his argument for determinism and concludes with some discussion on the implication that people are not the authors of their actions.  Since much of morality considers the intent to cause harm a prime consideration as to the morality of an act, does determinism make morality irrelevant.  Harris argues that it does not make morality irrelevant but rather it suggests that punishment does very little (except in a Pavlovian sense) to deter unwanted behaviour.  Determining desired and undesired behaviours (i.e. what is moral vs. immoral) does not change whether people have free will.  However, how justice is meted out must change once you accept that people do not have free will.  While containment of dangerous persons and deterrence through the threat of punishment is necessary for the greater good, rehabilitation should be the focus rather than incarceration as punishment.

The persistence of the illusion of free will is not difficult to explain.  Human beings tend to give themselves credit for good things and blame bad things on externalities.  When someone achieves a goal or completes a project or provides for their family or community, these good things are attributed as achievements of the self, including that they are self-generated, innovative, or even brilliant solutions to problems that others have failed to solve.  The credit is then given to the conscious self because of the physiological reaction of good feeling that follows.  Bad things that happen are blamed on externalities because if one admits the harm done, if often results in a physiological reaction of bad feeling.  Whether good achievements are self-generated or are the results of externalities does not diminish their good.  The key here about bad things is that humans do have a capacity to learn which can lead to a change in behaviour in a future similar situation.  This learning comes from the sum influences on the individual and is not a function of some illusory free will to learn.

References:

(0) Harris, Sam. Free Will. India, Free Press, 2012.

(1) https://www.thebigq.org/2020/11/26/what-are-some-of-the-key-developments-in-cognitive-neuroscience-%E2%96%B6/

(2) Ananthaswamy, Anil. The Man Who Wasn't There: Tales from the Edge of the Self. United States, Dutton, 2016.

(3) Judgment Under Uncertainty: Heuristics and Biases. Spain, Cambridge University Press, 1982.

(4) Asimov, Isaac. Foundation. United States, Random House Publishing Group, 2004.

(5) https://lifeinbeaverton.blogspot.com/2019/01/python-version-of-game-of-life-cellular.html