Daniel Dennett vs. Roger Penrose: Strong AI

 A Short

A while back, the philosopher Daniel Dennett made it seem as if the mathematical physicist Roger Penrose took an immediate and emotional position against strong artificial intelligence (sAI). Indeed Dennett quotes Penrose as saying: “Somehow I’ve got to prove that this is wrong.” In other words, Penrose had non-scientific reasons for rejecting sAI… But is that actually a quote from — or even a paraphrase of — Penrose? I say that because Dennett uses the phrase “he thought” rather than “he said”. This may well be a simple slip of grammar on Dennett’s part. Though if not, then Dennett was simply guessing as to what Penrose thought. After all, Dennett can’t have known what Penrose thought unless he told him.

In any case, the main position that Dennett upholds is that strong AI is actually entirely in tune with both physics and biology; whereas Penrose’s position seems to be at odds with these disciplines. In Dennett’s own words:

“[W]hat [Penrose] has seen… is that the only way you’re going to show that the idea of strong artificial intelligence is wrong is by overthrowing all of physics and most of biology!”

This seems to make intuitive sense. That is, it seems that nothing in sAI clashes with either physics or biology. What Penrose argues, on the other hand, does clash with both. But there’s a problem here for Dennett. Physics (as physics) has nothing to say about intelligence, let alone about consciousness or even life. Thus when AI theorists talk about intelligence, consciousness and life, they’re essentially going beyond physics. So in that sense, strong AI isn’t really in tune with physics at all. Of course it can now be said that nothing in sAI actually and clearly contradicts anything in physics. That may be true; though it is hard to decipher how any talk (by AI theorists) of intelligence, consciousness or life could contradict anything in physics.

What about sAI squaring with biology or natural selection?

Dennett is on safer ground here in the simple sense that some evolutionary theorists do indeed talk about intelligence and life. And some even talk about consciousness. (Other evolutionary theorists have “no need for the hypothesis of consciousness”.)

In any case, Dennett is correct to argue that a “revolution in physics” is required in order to sustain Penrose’s scepticism about sAI. He’s correct primarily because Penrose has himself often talked about the need for such a revolution. Primarily, that revolution is required to square relativity theory and quantum mechanics (see here). However, this revolution also ties in with the revolution that’s required to get to grips with the nature of consciousness — in which both quantum mechanics and gravity, according to Penrose, must play a role at the level of the brain's microtubules (see here).

Jaegwon Kim’s Epiphenomenalism?

 A Short

How do we solve the problem of epiphenomenalism? (This is a position summed up well in this way: “Subjective mental events are completely dependent for their existence on corresponding physical and biochemical events within the human body yet themselves have no causal efficacy on physical events”.)

Here’s one possible way offered by the Korean-American philosopher Jaegwon Kim.

Kim argues that “we should not think of the relation of neural events to their supervening mental events as causal”. Moreover,

“supervening mental events have no causal status apart from their supervenience on neurophysiological events that have ‘a more direct causal role’”.

This means that this isn’t a case of a mental event at time t causing a neurophysiological event at time t 1. No: if a mental event (M) supervenes on a neurophysiological event (N), then both the mental event and the neurophysiological event occur at one and the same time. So we can’t say that mental event M causes neural event N if they occur at one and the same time.

This is like the H₂0-water case.

That is, we don’t first have a set of H₂0 molecules which cause water. When we have a set of H₂0 molecules, we also have water. H₂0 molecules, then, don’t bring about (or cause) water. H₂0 molecules constitute (or are identical to) water.

On Jaegwon Kim’s account, we both can and cannot say that mental events are epiphenomenal. We can say that they’re epiphenomenal in that if mental event M occurs at one and the same time as N, then it can’t be epiphenomenal. More accurately, if M “inherits the causal power” of N (or if it “rides piggy back” on N), then it both has causal power and is not epiphenomenal. On the other hand, if M is indeed riding pigging back on N, then we can see it as being epiphenomenal. But, again, if M and N occur at one and the same time and are intimately connected, and if M also inherits the causal power of N, then in what sense is M truly epiphenomenal? (Perhaps this is simply a matter of taste or grammar.)

But if we tie M and N so closely together, aren’t we also saying (or implying) that M literally is N (as already hinted at with the H₂0-water example)? That is, that M and N are identical? Again, as with the epiphenomenal nature of M, both yes and no. From the outside, all we have is N and human behaviour. However, ontologically (or from the inside), M is clearly not identical to N. Again, from the first-person (or subjective) point of view, M is not a physical event. From the third-person point of view, on the other hand, M doesn’t factor at all except in terms of behaviour and “verbal reports”. M alone is also (according to the American philosopher Donald Davidson) incapable of falling under any natural laws.

References

Kim, Jaegwon, ‘Multiple realization and the metaphysics of reduction’ (1992).
 — ‘Mental Causation’ (chapter 6) in his Philosophy of Mind (1996).

Merely Verbal Ado About Nothing: David Chalmers, Facts, Consciousness

 

Contents: i) Introduction ii) The Facts and What We Say About Them iii) Stipulation Examples: iv) A Random Cup v) Is a Virus Alive? vi) First-Person Data vii) Eliminative Materialists vs. Reductive Functionalists viii) Causation ix) Bridge Laws x) Conclusion: Facts Matter and Consciousness

The title of this piece is partly based on David Chalmers’ paper ‘Verbal Disputes’. However, I relied far more on Chalmers’ book The Consciousness Mind: In Search of a Fundamental Theory than I did on his ‘Verbal Disputes’. (This paper is very long — 48 pages — and it covers many subjects.) Chalmers also gave a seminar on this subject; which can be found on YouTube here (see image above).

Many philosophers have tackled the “problem” of whether certain philosophical issues are “merely verbal disputes” or not. This debate goes back through the centuries. (Perhaps it was best highlighted by the logical positivists in the 1920s and 1930s.) In terms of contemporary philosophy, the notion that some philosophical issues are merely verbal has often been leveled at what is now called “analytic metaphysics”. Chalmers himself tackles some of these issues. (For example, he discusses whether a random booklike x is a book. Or, to use the contemporary jargon, is it only a collection of “particles arranged” bookwise?) However, Chalmers himself never names names and he certainly doesn’t use the term “analytic metaphysics”.

The Facts and What We Say About Them

Philosophy-scientist Smith has access to all the facts, laws, information, etc. about spatiotemporal slice (or state of affairs) A and says that it is x, y and z. Philosopher-scientist Jones has access to all the same facts about the same spatiotemporal slice (or state of affairs) A and says that it is a, b and c. Yet both Smith and Jones agree on the facts. This must mean that what Smith and Jones say about A is over and above the facts. In addition to facts, Smith and Jones needed to bring in theory, conceptual decisions, prior semantics, etc. into the discussion.

The given facts may well be determinate; though it doesn’t follow from this that what we say about them is also determinate. Or, in another manner of speaking, the facts alone don’t entail what we say about them…

But hang on a minute.

One may now wonder how this clean and neat distinction between facts and what we say about them can be upheld. After all, aren’t the facts (or what we take to be the facts) themselves somewhat dependent on what we say? David Chalmers himself doesn’t only argue that what we say is indeterminate. He also argues that “the facts [themselves] are indeterminate”.

Much of what’s just been said is fairly standard in science and in the philosophy of science. That is, the very same facts (or data) may engender different theories. Indeed some philosophers have argued that the very same facts (or data) could engender a (possible) infinite amount of theories. This situation is called the underdetermination of theory by the data and has been widely discussed in analytic philosophy.

And here again we can question the clean and neat separation of empirical data from the theories which, it seems to be supposed, come later.

Stipulation

Now Chalmers often mentions what he calls “stipulation”. The basic point is that if we stipulate what we mean by a particular word, then the answers to the questions about facts, data, what x is, etc. must — at least partly — follow from such stipulations. Of course some people will be horrified by the argument that acts of stipulation are decisive when it comes to what we take to be matters of fact. But it’s not that simple.

There is a problem with over-stressing the importance of stipulation; or even with simply emphasising the importance of stipulation at all. Chalmers sums up this problem with a joke. He writes:

“One might as well define ‘world peace’ as ‘a ham sandwich.’ Achieving world peace becomes much easier, but it is a hollow achievement.”

As it is, Chalmers only applies his joke to a single case: consciousness. So perhaps it can also be applied to other cases (such as the later cases of a random cup/book, virus, etc.). Clearly, even someone who argues that stipulation is important won’t also accept that we can define the words “world peace” as “a ham sandwich”. In turn, some philosophers and laypersons will feel just as strongly about claiming that, say, a “computer virus is alive” or that “bacteria learn”. The philosopher P.M.S. Hacker, for example, holds a very strong position on the philosophers and scientists who use such terms (or words) in ways that are radically at odds with everyday usage (see Hacker’s ‘Languages, Minds and Brain’ in Mindwaves). Many physicists, on the other hand, are very keen on using old words (or terms) in very different ways. (Think here of “information”, “space”, “time”, “intelligence”, “law”, “string”, “hole”, etc.)

Some Examples

A Random Cup and a Random Book

Let’s look at some more examples from David Chalmers. He asks:

“Is a cup-shaped object made of tissues a cup?”

The problem here is that it’s not clear if Chalmers meant toilet tissues or biological tissues in this example. In the former case, then, toilet tissues wouldn’t hold liquid. Thus, surely by definition, any x made out of toilet tissues couldn’t be a cup… Or could it?

What about biological tissues which could hold liquid?

In any case, let’s take it that whatever Chalmers meant by “tissues”, these tissues can indeed hold liquid.

Now take this question:

Is it the case that if any x functions as a cup, then surely it is a cup?

Here’s another question:

What if this cup-shaped (or particles arranged cupwise) object wasn’t designed to be cup?

Does that matter? If it holds liquid, and it even looks like a cup, then surely it is a cup. Why does it matter that it wasn’t designed to be a cup? Is whether it does or doesn’t matter a purely stipulative matter? In other words, is the following the case? -

x can only be classed as a “cup” if it were designed to be a cup.

This would mean that any natural object which were used as a cup could never be classed as a “cup” or even be a cup. Yet all sorts of natural things are used as functional devices which we then name according to their functions (e.g., a stick classed as a “weapon”, extracted venom classed as “poisons”, etc.). Does their natural status stop them from being named as functional devices (such as weapons)?

In any case, whether people call x a “cup” or not, they’re all still talking about the same x. Not only that: in the tissue-cup case, all people agree that it looks like a cup and can be used as a cup. The only difference, then, is what Chalmers calls “terminology”.

Here’s another question from Chalmers:

“Is a booklike entity that coagulated randomly into existence a book?”

This is like the infinite monkey theorem in which, after an infinite amount time in which an infinite amount of monkeys play with a typewriter, at least one of them will produce the complete works of Shakespeare. (In an infinite amount of time, surely an infinite amount of monkeys will produce the entire works of Shakespeare an infinite amount of times.) Such is the nature of the logical possibility which Chalmers is so keen on.

In any case, I presume that Chalmers doesn’t only mean book-shaped or arranged bookwise. Surely something that’s simply shaped like a book can’t be a book. That’s because, after all, it may not have any words in it. Then again, if stipulation rules, then why can’t shape alone be a necessary and sufficient condition for bookhood?

But let’s say that this random book does contain words. Not only that: it contains words which make sense. Here again we can ask the following question:

Is it relevant that this “booklike entity” is natural and wasn’t produced to be a book?

After all, if it looks like a book and contains grammatical sentences, a coherent story, etc., then surely it must be a book.

Is a Virus Alive?

David Chalmers tackles the case of whether or not a virus is alive. He writes:

“On theory might hold that a virus is alive, for instance, whereas another might hold that it is not, so the facts about life are not determined by the physical facts… the facts about life are indeterminate.”

One can’t read off from the facts alone whether or not something is alive. In other words, there’s more to being alive than the facts. To use Chalmers’ term, the facts may well be determinate; though what we say about them isn’t.

Let’s say that everyone agrees that a virus moves. Everyone may agree on its genetic structure. Etc. But it doesn’t follow from all these facts that everyone also agrees that the virus is alive. Some philosophers or scientists may see the virus as being a (biological) machine and therefore neither alive nor dead. (Though why would a virus’s machine-like nature automatically mean that it’s not alive?) To take another extreme interpretation. Some philosophers may even see the virus as being a simulation or simply a projection of our minds.

But what of a computer virus? Chalmers asks:

“Is a computer virus alive?”

This taps into the ancient debate of vitalism. Perhaps those same arguments should also be used to show that computer viruses are either alive or dead. It may be more complicated this time around; though functional, structural and physical criteria will be just as important as they were when vitalism was finally given up.

Indeed there’s a radical aspect to this. If the criteria of aliveness which worked for biological beings can also be applied to computer viruses, and also be acceptably or justifiably applied, then computer viruses must also be alive. After all, the term “computer virus” was coined precisely because such a thing fulfilled most — or even all — the functional, structural and physical criteria for aliveness.

First-Person Data

Chalmers gives an interesting example of the facts not determining what we say about them. It’s interesting because the status of these particular facts can itself be disputed. In addition, the subject area is one that’s rarely given as an example within this particular philosophical context.

Chalmers’ subject is what he calls “first-person data”. That is, what people say (or report) about their own conscious experiences or mental states. The basic point is that “all sorts of theories remain compatible” with such data,

“from solipsistic theories (in which only I am conscious) to panpsychist theories (in which everything is conscious); from biochemicalist theories (in which consciousness arises only from certain biochemical organizations) to computationalist theories (in which consciousness arises from anything with the right sort of computational organization); including along the way such bizarre theories as the theory that people are only conscious in odd-numbered years (right now, it is 1995)”.

The point here is:

“How can we rules out any of these theories, given that we cannot poke inside others’ minds to measure their conscious experience?”

What’s more, “[a]ll such theories are logically compatible with the data, but this is not enough to make them plausible”.

Now this particular example is problematic because it’s hard to see first-person data as being factual in the first place. That is, even if first-persona data consist in “verbal reports” which are indeed scientifically kosher, it’s still the case that the subject matter of those verbal reports may not itself be scientifically kosher. In any case, the facts alone (in this case) don’t necessitate what we say about them (i.e., our theories, concepts, words, statements, etc.).

But, here again, the facts/what we say about facts opposition can of course be questioned.

Still, what about the case when two people agree on the facts and yet say different things about them? That is, they don’t say different things about what the facts are and what their natures are. What they disagree on is what follows from the facts or how the facts are interpreted.

So, in Chalmers’ example, those in disagreement accept that subject S is having a mental state that, say, involves an experience of a red rose. They agree on this because they agree on S’s verbal reports about his own experiences or mental states. Now to get back to what Chalmers has already stated: that this very experience of a red rose can be explained in terms of a solipsistic theory, a panpsychist theory, a “biochemicalist” theory, a computationalist theory and an odd-numbered years theory. In other words, S’s experience of a red rose (not the red rose itself — if the two can be completely distinguished at all) isn’t doubted and even its nature may be agreed upon. The problem comes when that experience is theorised about — or interpreted — in different ways. In other words, there is an experience of a red rose (or, more correctly, the experience of a red rose is verbally reported); and it may even have a specific nature (despite it being first-person). However, how do we explain the experience itself? How do we account for it?

Eliminativist Materialists vs. Reductive Functionalists

Chalmers gives another example when he compares the positions of reductive functionalism and eliminative materialism. Here again the reductive functionalist and eliminative materialist both (more or less) agree on the facts. However, they still disagree on what Chalmers calls “terminology”. In this case, the eliminativist materialist and reductive functionalist (more or less) agree on the fact that “there is discrimination, categorization, accessibility, reportability, and the the like”. They even (more or less) agree on the philosophical and scientific accounts of such things. Therefore the only thing they disagree on (at least according to Chalmers) is that the reductive functionalist believes that “some of these explananda deserve the name ‘experience’”. The eliminativist materialist, on the other hand, believes that “none of them do”.

If discrimination, categorization, accessibility, reportability, etc. literally are — or literally constitute — experience, then surely experience (or simply the word “experience”) can be eliminated (at least in theory). In other words, experience (or the word “experience”) adds nothing to the pot. So this would mean that disagreement in this case truly is merely verbal.

Of course there may still be what’s called “semantic indeterminacy” when it comes to words like “discrimination”, “accessibility”, “reportability” and “categorization”. (Some philosophers have argued that this kind of semantic indeterminacy exists across the board. Others philosophers have also argued that it can’t exist across the board because such a state of affairs would somehow render communication — and even communal action — impossible.)

Causation

Chalmers even uses causation — or at least necessary causal relations — to highlight the point that theories, concepts, etc. are over and above the facts. This is (it can be supposed) Chalmers’ take on what’s called Humean supervenience (which has been much discussed in analytic philosophy).

Firstly, we have the facts about physical “regularities”. But what if “causation is construed as something over and above the presence of a regularity”? Indeed Chalmers goes so far as to say that “it is not clear that we can know that [causation] exists”.

To be clear, this isn’t really about the strong distinctions which can be made between the many things which supervene and their “supervenience bases”. It’s about the actual “failure of logical supervenience”. More explicitly:

“[F]acts about causation fail to supervene logically on matters of particular physical fact.”

Thus anything we say about the causation doesn’t “logical supervene” on the facts alone. So even causation (like our stipulations about what a cup/book is, what is alive, etc.) are over and above the facts. Does this mean that causation-talk too is merely verbal?

All this depends on what Chalmers means by “causation”.

The 18th century philosopher David Hume would have accepted that B always follows A. However, there’s no necessary link between the two that’s somehow over and above what we observe. Thus if there’s no necessary link, then B simply follows A. After all, Chalmers himself distinguishes causation from “mere succession”. But does this Humean picture automatically mean that we don’t actually have causation at all? Is non-observable (or non-empirical) metaphysical necessity necessarily built into all talk of causation?

In any case, this isn’t really the place to discuss Humean supervenience or even causation. The point is, though, that whatever philosophical position we take on “mere succession” (or causal relations) it will be over and above what it is that’s “behind”, “beneath” or “between” the successions (or relations) we talk about. Basically, what we say about physical (or causal) relations is (or can be) over and above the facts.

Bridge Laws

Philosophers (specifically in the philosophy of mind) often use the technical term bridge laws. Bridge laws are said to tie lower-level phenomena to higher-level phenomena. In the case of the philosophy of mind, facts (since that word has been used a lot in this piece) about the brain are tied to things (not facts) about the mind, experience or consciousness.

Chalmers argues that bridge laws are over and above the facts. This is Chalmers’ own take on bridge laws:

“Some might argue that explanation of any high-level phenomena will postulate ‘bridge laws’ in addition to a low-level account, and that it is only with the aid of these bridge laws that the details of the high-level phenomena are derived.”

Chalmers suggests (or states) that “in such cases the bridge laws are not further facts about the world”. That is, “the connecting principles themselves are logically supervenient on the low-level facts”. In other words, these connecting principles are not facts. (Alternatively, the statements about connecting principles aren’t factual.) Chalmers then gives an obvious and clear example of this: “the link between molecular motion and heat”. Heat simply is what’s called “mean molecular motion”. (Or: heat = mean molecular motion.) Having said that, there are things which can be said about heat which can’t be said about moving molecules. All talk of heat, nonetheless, can still “be derived from the physical facts”. Still, things said about heat are over and above the things said about (mean) molecular motion. What’s more, what’s said about heat doesn’t include “further facts about the world”.

This raises the question:

If not further facts about the world, then further… what?

Chalmers trumps all this fairly uncontroversial stuff with — as one might have guessed — an exception to his general rule: consciousness. In the case of consciousness (so Chalmers believes), consciousness is a “further fact[] about the world”. What’s more, consciousness is not (again) “logically supervenient on the low-level facts”. Consciousness may be empirically and contingently supervenient on low-level facts; though consciousness isn’t logically supervenient on them. That is, no physical facts about the brain (or otherwise) logically entail consciousness; and consciousness doesn’t logically entail any facts about the brain (or otherwise).

*********************************

Facts Matter: A Mouse’s Beliefs and its Conscious Experiences

Here’s another question from Chalmers:

“Does a mouse have beliefs?”

As stated in the introduction, Chalmers often mentions “stipulation”. That is, if we stipulate what we mean by the word “belief”, then the answer to that question must — at least in part — follow from the stipulation.

To simplify, if x, y and z constitute what it is for something to be a belief, then if a mouse displays x, y and z, then it has a belief. This is of course a simplified story. That’s because agreement will have to be made on x, y and z, and then on whether not x, y and z are necessary and sufficient for belief. But however complicated this story turns out, stipulation will still remain part of it. That is, do we believe that (as it were) beliefness (like aliveness) is something over and above the functional, structural and/or physical facts?

Chalmers is keen to accept the importance of stipulation when it comes to such decisions. He also believes — at least as I see it — that much that passes for metaphysics is merely verbal dispute. However, it’s still the case that in some cases (or in one case!) at least there’s a fact of the matter which makes some statements, concepts or theories plain wrong.

Take Chalmers’ own final question:

“Does a mouse have conscious experience?”

In this case, it isn’t all about stipulation or verbal dispute. That is:

“Either there is something that it is like to be a mouse or there is not, and it is not up to us to define the mouse’s experience into or out of existence.”

So it’s not always a case of all the debaters agreeing on the facts; though still disagreeing on what they say about the facts. This time — at least according to Chalmers — the debaters are also disagreeing about the facts. In this example, it’s about whether or not “a mouse [actually has] conscious experience”.

In the previous examples the debaters said different things about the facts; but agreed on the facts. Now the debaters disagree on the actual facts. What’s more, Chalmers believes that “we cannot stipulate [] away” whether or not the mouse has conscious experience or not.

The question is, then, whether or not Chalmers’ position on consciousness really is in a different ballpark to the previous disputes about computer viruses, mice having beliefs, books made out of tissues, bacteria which learn, etc. That is, is a “functional analysis” also acceptable in the case of a mouse’s experiences? Chalmers says “no”.

So my own final question is:

Why is the case of a mouse having — or not having — “conscious experience” so different to the cases already discussed?

‘Tenet’: Does Christopher Nolan’s Latest Film Make Scientific Sense?

 

Despite all the nitpicking which will follow, there’s absolutely no reason why Christopher Nolan’s scientific name droppings and speculations (in his film Tenet) should be scientifically accurate in every respect. (It can even be argued that they needn’t be accurate at all.) Indeed, Nolan has himself said that he took various liberties in Tenet. (See his later reference to Kip Thorne.)

But what else would anyone expect?

Tenet is a work of art/entertainment, not a piece of scientific research.

So there are many scientific problems with — as well as questions about — Tenet. However, to grind through each one of them would be to verge on both scientific and aesthetic pedantry. And that’s why only the main scientific themes of this film have been tackled in this piece. Namely: entropy, time travel (specifically, the Grandfather paradox) and “parallel worlds”.

************************************

Tenet is a new spy film. It’s directed and written by Christopher Nolan, who previously directed The Dark Knight Trilogy (2005–2012), Dunkirk, etc. More relevantly to the theme of the science in science fiction, Nolan also directed the sci-fi movies Inception and Interstellar.

Nolan took more than five years to write the screenplay for Tenet. Not only that: he worked on some suggestions from the theoretical physicist Kip Thorne. (Kip Thorne had previously offered advice on Nolan’s film Interstellar.) More relevantly, Thorne (in his Black Holes & Time Warps) wrote about going down a wormhole in which

“within a fraction of your second of your own time you will arrive on Earth, in the era of your youth 4 billion years ago”.

Christopher Nolan himself has said that

“Kip Thorne read the script and he helped me out with some of the concepts, though we’re not going to make any case for this being scientifically accurate”.

 

Amateur Film Review

The basic gist of Tenet is that a secret agent called the Protagonist (played by John David Washington) manipulates time (via time travel or “inverted entropy”) to prevent a global catastrophe.

In terms of film criticism, I must disagree with the critic who wrote the following:

“A visually dazzling puzzle for film lovers to unlock, Tenet serves up all the cerebral spectacle audiences expect from a Christopher Nolan production.”

I don’t believe that Tenet is particularly “cerebral”. Apart from the science arguably being weak or at least exaggerated (if that matters at all in a Hollywood film), multiple things happening at once doesn’t necessarily mean that something is cerebral. And I don’t agree with film critic James Berardinelli either when he wrote that Tenet

“may be the most challenging of Nolan’s films to date when it comes to wrapping one’s mind around the concepts forming the narrative’s foundation: backwards-moving entropy, non-linear thinking, temporal paradoxes”.

Although we have a host of technical terms from physics (specifically, the more popular sexy ones) in Tenet, none of these is developed in any way. (There is no strong — aesthetic — reason why they should be developed.) Robert Pattinson’s character, Neil (the Protagonist’s “handler”), for example, mentions the positron. He then mentions Richard Feynman on the backwards time travel of the positron. This is a reference to Feynman’s interpretation — repeat, interpretation (see here) — that the positron as an electron moving backward in time. To state the obvious, there’s a huge leap from this interpretation of a positron (as a electron moving backward in time) to bullets - let alone human beings - moving backward in time. For a start, we’ve moved from the “quantum scale” to the “classical scale”. Having said that, there’s much dispute in science and philosophy as to whether there is such a neat and clean “line in the sand” (an image from the physicist and science writer Brian Greene) between the quantum and classical realms.

Some Scientific Parts of the Plot

Andrei Sator, the Russian oligarch (played by Kenneth Branagh).

In terms of the plot, the Protagonist is led toward a secrete organisation called Tenet. (The word “tenet” itself can be inverted.) In that organisation a scientist called Laura is studying bullets whose entropy has been “inverted”. Such inversion enables bullets to move backward through time. And, of course, it turns out that people can be temporally inverted too.

The plot is thickened by the fact that these counter-entropic bullets have been used by a Russian oligarch by the name of Andrei Sator (played by Kenneth Branagh). And Sator is himself communicating with future persons.

Technically, these inversions are carried out by a machine called a “Turnstile” (which is like an up-to-date version of H.G. Well’s “time machine” or Doctor Who’s TARDIS). The Turnstile itself was created in the future.

In terms of detail, various inverted devices (other than bullets) are said to be the products of a “algorithm” which was also developed in the future. That algorithm is capable of inverting the entire world. Indeed future humans are using Sator — knowing that he’s “mad” enough to do it — to activate it.

Entropy and the Science of Time Travel

Time travel is certainly seen as a possibility by many scientists. Albert Einstein, for example, was keen on this debate and offered his own ideas on the possibility. He even showed us how he believed it could be done. And, many times, Stephen Hawking too entered the arena of time travel.

Despite all that, many other scientists reject time travel. Some of their rejections are purely logical/philosophical. Others are purely scientific. Yet more are both scientific and philosophical/logical.

In terms of possibility, we also have such cases as wormholes (as Kip Thorne mentioned), Kurt Gödel’s rotating universe, black holes, etc.

As for entropy.

At first it will be hard for the average viewer to to see the precise connection between entropy and time travel in the film Tenet. One may understand entropy fairly well. One may also understand time travel fairly well. However, exactly how are these two things tied together in Tenet?

One important point to get out of the way is that in physics all physical processes (or events) are deemed to be time symmetrical. That is, if one plays the film back of a physical event, then such a occurrence would be physically possible. What’s more, the viewer wouldn’t be able to spot the difference between something being played back and something being played forward. However, this is true only of processes (or events) in the quantum realm, not the “classical” realm. In other words, the processes or events involving subatomic particles, their interactions, etc. within a specific experiment (or within a given sample space) can be reversed. However, if a football match were to be reversed, then things would be very different. That is, although a viewer would quickly note the time reversal of a football match, and if that reversed match were real, it still wouldn’t go against any of the laws of physics.

To get back to entropy.

Of course it can be said that the “inversion” of entropy is very different to backward or forward time travel (despite the fact that they’re tightly connected in Tenet). More specifically, rather than increased entropy occurring within the universe, the opposite can occur in any given system. That is, system S can become more ordered over time. But this isn’t so surprising. Many systems become more ordered over time despite the overall entropy of the universe.

This means that entropy applies to the universe as a whole, though not to, say, a bullet (one of the first examples in Tenet).

Of course the fridge is the best-known example of what’s sometimes called “reverse entropy” (“negative entropy”). And when the fridge first became generally available, it even sparked the interest of Albert Einstein. (He applied for a patent.) However, there are no conclusive reasons (as yet) to believe that this is also a question of the the reversal (or inversion) of time.

The problem (or simple fact) is that although a fridge does reverse entropy, it only does so in a “local” and insulted environment (i.e., within an enclosed system). In other words, a refrigerator alone — and even every fridge on the planet — can’t reverse the entropy of the universe. And neither does a single fridge - even in its own local space — make time run backwards.

On a more technical “microscale” than a fridge, scientists have used a strong magnetic field to make the nuclei in hydrogen particles of chloroform get hotter, while their colder carbon partners get colder (see here).

Yet here again there is no reversal of time. (This, of course, depends on what time is taken to be.)

So how does inverting entropy automatically entail going backward in time? And why should all this apply to both going backward in time and going forward in time?

Parallel Worlds

Logically (rather than scientifically) one can conceive of parallel worlds (since such a thing is mentioned in Tenet) in which entropy decreases rather than increases. The important word here is “conceive”. This hints at the fact that this is a philosophical/logical scenario rather than a scientific one (though some/many scientists may disagree). Not only that: since possible worlds (at least in philosophy) are primarily about what’s often called logical possibility (rather than scientific/empirical possibility), this further problematises this issue.

The Tenet character Neil also mentions “parallel worlds”. Here it’s worth making an important distinction (for the layperson) between other dimensions and parallel worlds (or Hugh Everett’s many worlds). Other dimensions are dimensions of our universe. (This is a fundamental part of string theory.) Parallel worlds, on the other hand, may be completely separate from our word. That is, can there really be “causal contact” between parallel worlds and our world? This must surely mean that the only possibility that Tenet can have in mind is parallel dimensions, not parallel worlds. Indeed when it comes to parallel worlds, it’s hard to make sense of the word “parallel” at all. These worlds simply can’t be parallel to our world. (Of course perhaps no one is expected to take the word “parallel” literally.)

And just as a distinction has just been made between parallel dimensions and parallel worlds, so a distinction can also be made between the many worlds (of many-worlds theory) and the other “universes” of a multiverse. In many-worlds theory, the possibility of causal — or other kinds of — contact between our world and other worlds is accepted — at least by some scientists. (See also brane cosmology.) However, when it comes to our universe and other universes (or a “bubble universe”) in a multiverse, that isn’t usually accepted by scientists (although there is some talk of “colliding” universes).

The Grandfather Paradox

One of the main characters in Tenet mentions the Grandfather paradox. This has it (in basic terms) that a person goes backward in time to kill his own grandfather. Yet this act seemingly makes it impossible for him to do such a thing in the first place. That’s because, were he to be successful, then he wouldn’t exist. In other words, it would never have been possible for him to go back and kill his own grandfather had his grandfather already been killed (i.e., by this person as a time-traveler).

The Protagonist asks which paradoxical possibility is true when it comes to the Grandfather paradox. Neil (the Protagonist’s handler) answers by saying that none is true because “that’s the nature of paradoxes”. This seems like a cheap and easy way to bypass the philosophical — and, perhaps more relevantly, scientific — problems with this aspect of time travel.

In Tenet, however, we have much more than mere backward or even forward time travel. We also have all these things happening at the same “time”. Not only that: the same person exists at the same time as… well, two people. So Tenet presents us with multiple juxtapositions of time travel. Take these:

1) Moving from (our) present to the future. 
2) Moving from (our) present to the past. 
3) Moving from the future to (our) present. 
4) Moving from the future to (our) past. 
5) Moving from the past to (our) present. 
6) Moving from the past to (our) future.

And no doubt I’ve left some options out here.

Thus the Protagonist — as he exists in the present — is in the same scene (or time) as the Protagonist as someone who’s traveled back to the present from the future. However, since all time must be relative in this film, it must be difficult for the viewer — and the world itself - to tell what is the present.

One other example (of many) in which the same person exists as two people in the same scene is when Sator (the Russian oligarch) arrives at a location only to see his inversion (i.e., himself) leaving that very same location.

One final example of different types of time travel occurring at the same “time” occurs when troops carry out a “temporal pincer movement” in which one set of troops moves forward in time and another set moves backward in time. It’s difficult, however, to understand both what this means and how it’s possible. In terms of the visuals of Tenet itself, this is simply backward and forward movement in the same visual scene. Yet in this sci-fi scenario we’re supposed to have backward and forward time-travelling occurring in the same place and at the same “time” — at least as it’s presented to the viewer. Thus, in that case, wouldn’t we have three spacetimes at the same… well, time? That is, in the present time and place we’d have one set of troops moving backward in time and another set of troops moving forward in time. In addition, all this would be occurring in the same phenomenological (as it were) “timeless present” (i.e., from the point of view of the viewer).

It can be supposed that we could have different things happening in different spacetime slices of the same overall spacetime totality. That is, within spacetime area A, we’d also have spacetime area x and spacetime area y. (Having put that possibility, I doubt that this was the conscious intention of the Christopher Nolan or of anyone else.)

Ironically, despite all these attempts to change the past, the character Neil says (more than once) that “what happened, happened”. This seems to be a hint (or a plain statement) that the past simply can’t be changed..

Of course one solution to all this is not to argue that the past can be changed or that it can’t be changed. Instead, event x happened in the past and a time traveler goes back to just before event x in order to stop it from happening in the first place. However, when he does stop it from happening, what actually happens is that event x still happens in our world; though it doesn’t happen in a “parallel world”… But what’s the point of that? This still means that event x occurred in our world. And, as many philosophers and scientists have said, what do these other worlds have to do with us?