“Physical Time within Human Time” and “Bridging the Neuroscience and Physics of Time”

In: Timing & Time Perception
Yuval Dolev Department of Philosophy, Bar-Ilan University, Ramat Gan, 5290002, Israel

Search for other papers by Yuval Dolev in
Current site
Google Scholar
Open Access


The commentary discusses whether AI devices can have an experience of time as passing. The papers commented on suggest the answer is “Yes”. However, I claim that the metaphysical view of time this answer presupposes, namely, eternalism, or the block universe, is untenable, and that a sound understanding of time must acknowledge time’s normative significance. This raises new and substantial questions regarding the possibility of representing time, and of equipping devices with “gadgets” that would facilitate as-of flow experiences for them. More generally, the commentary critically evaluates the role science may have in the context of philosophical debates.

In their paper, Buonomano and Rovelli (2023) state, among other things, that “relativity does not permit an objective notion of a global present; that the distinction between past and future requires thermodynamics, hence is statistical only”; and that “the arrow of time is not part of the basic grammar of nature”. According to Buonomano and Rovelli, flow and “the subjective ‘now’”, alongside “our experience of being ‘free to choose’”, “emerge from the brain”. Objections to the claim that physical time is flowless and tenseless are preempted by the standard analogy: flow and tense are like the flatness of earth — they hold locally, that is, they are “products of the brain’s architecture”, which is “adapted to our local environment”. To regard them as constitutive of time is to make the anti-scientific mistake committed by flat-earthers, who infer from local, subjective experience to the fundamental structure of reality (blurring the line between ‘local’ and ‘subjective’ is commonplace). This position has startling consequences, for example, that it is meaningless to ask what an astronaut orbiting around Saturn is doing now — the ‘now’ does not extend that far; and that temporal direction, which “is caused by low entropy in the early universe”, is a matter of empirical contingency, with the corollary that “it is far from obvious why we remember the past but not the future”.

This doctrine is well-known and popular, yet I am convinced (and have argued in many publications), incoherent. For one thing, the analogy with the local flatness of earth does not hold water. Spatial local flatness has a precise geometrical meaning; in contrast, a local ‘now’ is a Pandora’s box of conceptual muddles. Here’s what we know from experience and language: to be an event is to be in time, and to be in time is to be either past, or present or future, regardless of the event’s location in space or on the calendar. It is events in the physical world themselves, not events as they are experienced, that are subject to flow and tense. Tense and flow are not detachable add-ons events get ornamented with only in how they are experienced. To experience, speak, and think of an event, whether you are a neuroscientist, physicist, or ordinary person, is invariably and irremovably to think of it as being either past, or present or future. This is true also of all scientific contexts. It’s not as if anyone has a tenseless, flowless language to resort to as an alternative. It follows (from arguments which cannot be elaborated here due to space constraints) that no coherent sense can be attached to the idea that we “remember the future”, or that events on Alpha Centauri are neither past, nor present nor future. If you can question whether memories are necessarily of the past, you might as well question everything, including the meanings of the very words with which you do science, and including the meanings of the very words with which you articulate your doubt, which is why skepticism of this kind is self-defeating.

Disturbingly, the view that flow is an illusion and that tense is subjective and local is often put forth as ‘established science’, including in the Buonomano and Rovelli (2023) paper. But can science settle the age-old philosophical question regarding the reality of flow, and decide which view is correct — the one that takes flow to be the essence of objective time or the one that banishes it from reality altogether and relegates it to psychology? If, as Einstein claimed, “physics has no possibility of expression for ‘now’ (present), for ‘past’ and for ‘future’” (see Note 1), the answer is ‘no’. If physics cannot speak of flow, then it is unable to say anything of flow, specifically, it cannot deny nor confirm its reality, and it cannot deliver empirical confirmation either that flow is an illusion or that it is not. But then it follows that when scientists state that tense and/or flow are an illusion, they are expressing a metaphysical opinion, not a bit of ‘established science’.

Of the elements that make up the arguments grounding these claims (arguments I have presented in detail elsewhere), I wish in this short note to stress two considerations.

First, the arguments purporting to show that time is tenseless and flowless are invariably circular — first a tenseless picture and a tenseless language are assumed, and then it is shown that reality is tenseless. To illustrate, Putnam’s (1967) argument, a pioneer among arguments establishing the tenseless view on the basis of relativity, is grounded in the far-from-obvious metaphysical (as opposed to scientific) “There are no privileged observers” premise: “If it is the case that all and only the things that stand in a certain relation R to me-now are real, and you-now are also real, then it is also the case that all and only the things that stand in the relation R to you-now are real”. In explaining the premise, Putnam adds that “R must be restricted to physical relations that are supposed to be independent of the choice of a coordinate system (as simultaneity was in classical physics) and to be definable in a ‘tenseless’ way in terms of the fundamental notions of physics” (1967, p. 241). First a relation crucial to the argument is defined as tenseless, and then a tenseless view is derived. In subsequent relativity-based arguments this kind of circularity is better concealed, but it plagues them, nonetheless.

Recently, information gathering and utilizing systems (IGUSs) were thrown into the fray and might be thought of as offering a new kind of scientific support for the view that tense and passage are only a ‘second-level’, emergent phenomenon in a tenseless world. If it can be shown that tenseless input can generate in IGUSs, robots. or humans, ‘representational states’ as-of tense and flow — if such devices can have experiences of flowing time, of a ‘now’, even though the universe is tenseless, then there is no reason to insist that flow and tense are actual and objective. But here again, the splitting of time into a ‘manifest’, experienced, psychological time and a tenseless block universe, is assumed. That’s not surprising — the only way to defend a view that goes against language and experience is by means of an alternative language.

The second observation is that flow and tense do not lend themselves to being represented. Dreyfus is famous for making the case, on the basis of the works of Heidegger and Merleau-Ponty, that our dealings with our environment are not mediated by, or dependent on, or facilitated by internal models of it. “The meaningful objects … among which we live are not a model of the world stored in our mind or brain; they are the world itself” (Dreyfus, 2007, p. 249). “Heidegger and Merleau-Ponty would say”, he conjectures, “that what AI researchers have to face and understand is [not onl]y why our everyday coping couldn’t be understood in terms of inferences from symbolic representations …” (p. 251). In walking out of a room, no ‘representational-state’ that models the door and the room is involved. The successful action of the agent is not the result of computations the brain performs on some internal proxy of the door.

Dreyfus’s claims and reasoning cannot be elaborated here. But happily, and surprisingly, perhaps, temporal flow constitutes an easier case for discussing the Dreyfusean outlook. Unlike tenseless relations, flow simply can’t be symbolically represented at all. There’s nothing in the mathematical formalism of physics that corresponds to or captures flow, no diagram that depicts it. You cannot put a number on it, or put it into numbers. For one thing, say you have a complete description of a system, a list specifying in full its state at any given moment, and you want to add what state the system is in now. How would you do that? Any marking of the present will be useless because the minute you make the mark, the state marked is no longer present. You can highlight the present state of the system ‘from the outside’, with a laser pointer, for example. But the red dot is external to the list of states, or graph depicting them. A dynamic representation, an animation, would be useless too — the only way to mark the present would be by reference, again, from the outside, to the part of the animation you are seeing now.

The experiential, even existential significance of flow cannot be overstated. Our most crucial sentiments, emotions and normative stances are utterly dependent on where in tensed time the events that make up our world and lives are situated — in the past, the present or the future. But flow and tense are not quantifiable, mathematizable. They are real and objective but they are given only through their experiential manifestations, and since these are inextricably tied to the normative significance of what we experience, they remain outside the scope of what can be symbolically representable.

As we learn from works such as Gruber et al.’s (2022), fully functional IGUSs, that is, systems that can perform every physical task we humans engage in, may be a possibility. And it is possible that the capacity of IGUSs to have illusory as-of flow ‘experiences’ is vital for their success in fulfilling various tasks. But they are representational creatures, whose every act is mediated by an internal mini-model of their environment, while we are simply in the world, with no interface, no intermediary, between us and it. And this monumental difference is clearly and loudly expressed in what we care about. Even if behaviorally IGUSs are very much like humans, they lack the main, constitutive characteristic of humans: normative sensitivity and sensibility. IGUSs don’t care about anything, they just are. Perhaps, if they can be equipped with a ‘flow gadget’, they can also be equipped with a ‘care gadget’, a ‘beauty gadget’, and so on. That will not change a thing — what for us are vitally important constituents of our environment, e.g., the well-being of other persons, are for them mere currents in a circuit. A full and accurate depiction of many of our actions will inescapably include reference to motives and concerns that drive us. Not so with IGUSs. Only if we could draw a wedge between our physical actions and their normative significance would we be able to say that we and IGUSs are the same. But we can’t.

The past decades have witnessed the appearance of hitherto unimaginable machines with capacities to emulate human speech, creativity, and strategizing. Possibly, the working hypothesis that human-like machines are possible is vital for this progress. The notion of IGUSs may be highly fruitful, as evidenced in papers such as Gruber et al.’s (2022). In view of such investigations, philosophers of humanistic convictions, who are staunchly committed to there being a fundamental and non-effaceable distinction between human experience and machine functioning, are forced to probe more deeply into this distinction and come up with new defenses of their commitment. There are substantial benefits to this effort, a better appreciation of the unrepresentable nature of time being one of them. But for those of the opposite persuasion, who believe that the scope of science includes everything and anything, that, at the end of the day, everything can be expressed by the dynamical equations of physics, the normativity challenge is as steep as it ever was, and exposes certain claims made in the name of ‘established science’ as, in fact, philosophical views, which, like every interesting position in philosophy, face significant difficulties.

The sustained effort to create human-like artificial devices continues to yield new technological breakthroughs and models for cognitive processes. But it does not blur the distinction between machines and persons, a distinction that also separates science from philosophy. Both sides can only benefit from remaining cognizant of this divide.



In a 1952 letter to Ruth Levitova.


  • Buonomano, D., & Rovelli, C. (2023). Bridging the neuroscience and physics of time. In R. Lestienne & P. A. Harris (Eds), Time and science, Vol. 2 (pp. 267282). New Jersey, NJ, USA: World Scientific.

    • Search Google Scholar
    • Export Citation
  • Dreyfus, H. L. (2007). Why Heideggerian AI failed and how fixing it would require making it more Heideggerian. Philos. Psychol., 20, 247268.

    • Search Google Scholar
    • Export Citation
  • Gruber, R. P., Block, R. A., & Montemayor, C. (2022). Physical time within human time, Front. Psychol., 13, 718505. doi: 10.3389/fpsyg.2022.718505.

    • Search Google Scholar
    • Export Citation
  • Putnam, H. (1967). Time and physical geometry. J. Philos., 64, 240247.

Content Metrics

All Time Past Year Past 30 Days
Abstract Views 38 38 0
Full Text Views 56 56 12
PDF Views & Downloads 102 102 30