Motion Perception and the Temporal Metaphysics of Consciousness
Henry Pollock and Samantha Strong
Abstract: This paper defends a ‘punctivist’ conception of consciousness from recent attacks by Ian Phillips and Matthew Soteriou. As we intend it, ‘punctivism’ is the view that a subject’s experience over some interval is determined by their experiential states at each instant during it. Phillips and Soteriou both offer ingenious arguments purporting to show that the punctivist is unable to make sense of motion perception; and that only by adopting an ‘holistic’ conception — whereby a subject’s instantaneous experiences are determined by their experience over the interval — can we make sense of the puzzles such phenomena pose. We contend that both arguments invoke dubious claims, their proffered solutions come with highly controversial commitments, and, if we take punctivism seriously, it is difficult to see why the so-called puzzles should be puzzling in the first place. A punctivist model of motion perception is proposed in response, and objections anticipated.
How does the mind fill time? This broad question is one to which two opposing conceptions of consciousness will provide very different answers. On one conception — what we might call a punctivist conception — the mental life of a subject over an interval is always determined by, and so can be completely explained in terms of, that subject’s mental states at each instant during that interval. The most basic building blocks of consciousness are on this view instantaneous. According to the other view — what we might call an holistic conception — the punctivist has it the wrong way round. Instantaneous mental states are attributable only in a derivative sense: a subject’s mental state at an instant is determined by what they are conscious of over some interval within which that instant falls. Extended experiences are metaphysically prior to instantaneous ones.
The punctivist conception has recently come under attack from Ian Phillips (2011) and Matthew Soteriou (2013). Both offer ingenious arguments purporting to show that the punctivist cannot make sense of motion perception, and that only by adopting the holistic conception are we able to solve the puzzles such experiences pose. This paper constitutes a counter-argument. Both arguments invoke dubious claims, their proffered solutions come with highly controversial commitments, and, if we take punctivism seriously, it is difficult to see why the so-called puzzles should be puzzling in the first place. There may of course be other considerations that support either conception. To be clear, our purpose is simply to show that holism should not be favoured for the reasons given by these two prominent holists, and that it is in fact the punctivist who is best placed to make sense of the phenomena in question. A punctivist model of motion perception is proposed, and objections anticipated.
2. The Two Conceptions
One may wonder why we have chosen ‘punctivism’ over ‘atomism’ as the name for the opposing theory to holism. The reason for this choice is simply that the literature on temporal experience is something of a terminological quagmire; and one in which ‘atomism’ is already inconsistently used between authors. Because of this, it will be beneficial to first say a little more about our terminology and how it fits into the wider debate.
Much of the philosophy of temporal experience is framed in terms of a debate between ‘extensional’, ‘retentional’, and ‘snapshot’ theorists. Extensionalists posit temporally extended experiences within which temporal phenomena unfold. The temporal structure of the extended experience more-or-less corresponds to what the experience is of. Retentionalists and snapshot theorists deny this, typically positing instantaneous experiences. But whereas the retentionalist holds that each individual experience represents some extended period, the snapshot theorist holds that our experience over time is ‘built up’ from successive ‘snapshots’ of instantaneous states of affairs.
Names for these models vary between writers, and sometimes vary even between publications from the same writer. This is where one finds the name ‘atomism’ cropping up. Philippe Chuard (2011), for example, uses the name ‘atomism’ in reference to his snapshot model. His ‘atomism’ is close to our ‘punctivism’, since he apparently holds that instantaneous experiences are fundamental. The two are not equivalent, however, since for us a punctivist is anyone who takes instantaneous experiences as fundamental — not just snapshot theorists.
Geoffrey Lee’s (2014) ‘atomism’, in contrast, is plausibly interpreted as a version of retentionalism. For Lee, experience is atomistic in the sense that it consists of fundamental units — atoms — which do not themselves have shorter experiences as proper temporal parts. This does not preclude the possibility that these atoms are temporally extended, though. As it happens, Lee contends that experiential atoms are indeed extended, but that they are instantiated holistically. On his view, although we can say that I am in pain at a time, this is only in virtue of my having an extended atomic experience of pain over that time. Lee’s atomism therefore turns out, by the definitions employed in this paper, to align with the holistic conception: for Lee, the fundamental units of experience are extended. If Lee was to espouse a more traditional retentionalism whereby experiential atoms are unextended, he would be a punctivist in our books.
Some might find it obscure how experiences could be fundamentally extended as the holist claims. By way of analogy, holists sometimes appeal to temporally extended activities. It is, for example, impossible for a person to walk for only an instant. Nevertheless, provided a person is walking over some interval, it will be true that they are walking at each instant during that interval (Soteriou, 2007). And so, it is claimed, the extended act of walking is metaphysically prior to the action-stages it comprises; the extended item, rather than its instantaneous parts, is fundamental.
One might wonder whether such analogies establish quite as much as the holists want them to. Simon Prosser (2016) contends that all they show is that an instantaneous part cannot exist in isolation. But this dependency fails to establish any metaphysical priority:
Suppose that an A cannot be ф unless it is followed by a B. This does not show that the A and the B together form a whole, or that this whole is more fundamental than its proper parts. For example, no one can be the second person to set foot in Antarctica unless there was a first person to set foot in Antarctica, but this does not show that the first and second persons form a whole. (ibid., p. 151)
Prosser’s example may not be entirely fair. It certainly would be odd to suppose that the first and second persons together form a fundamental whole. But the dependency in the Antarctica example doesn’t concern persons per se; it concerns their positions. The point would be that the position ‘second’ cannot be the position ‘second’ without being part of an ordering. So the dependency in Prosser’s example might indeed be thought to establish a fundamental whole — just not the one he attacks. It is, of course, another question whether consciousness is, in the relevant sense, like orderings. Is it metaphysically possible for an instantaneous experience to exist in isolation? Soteriou and Phillips think not.
3.1. Soteriou’s argument
Soteriou is a holist, since he argues that a subject’s mental state at a time can be determined by their mental state over time (2013, p. 106). He presents the holistic conception as the solution to an apparent puzzle concerning our perception of motion. The puzzle is generated by the following two assumptions:
(1) A subject’s awareness of a temporally extended occurrence will have the same temporal extension as the occurrence.
(2) ‘[I]f a subject is in a given state over a given period of time, then this is determined by… the fact that she is in that perceptual state at each of the instants that make up that period’ (ibid., p. 96).
Soteriou says that assumption (1) ‘might be motivated by the temporal transparency of experience’ (ibid., p. 97). Experience is said to be ‘transparent’ in the sense that when we introspect, we cannot discover any features of our experience other than those the experience is of. As Moore writes:
That which makes the sensation of blue a mental fact seems to escape us; it seems, if I may use a metaphor, to be transparent — we look through it and see nothing but the blue. (Moore, 1903, p. 446)
Regarding the temporal features of experience, this amounts to the following: we are unable to identify any temporal location or duration of the experience other than what is presented by the experience. In other words, one’s experience seems to run concurrently with its object. So when I am aware of an object moving from location L1 at time t1 to L10 at t10, it seems that my awareness of this motion occupies that same interval — viz. t1 to t10. For example, when I observe the second hand of a clock sweeping from five past to ten past, it seems as though my experience occurs as the second hand makes its sweep; and it seems as though my experience lasts for the five seconds the sweep does.
When a subject is aware of an occurrence, Soteriou supposes, they are in a state of awareness (2013, p. 94). So by (1), my awareness of the object’s motion from L1 to L10 spans some interval; and throughout this interval I am in a state of awareness of the object moving from L1 to L10.
The problem is that this is in tension with claim (2). According to (2), the fact that I am in a state of awareness of the object’s movement from L1 to L10 is determined by my being in that state of awareness at each of the instants during the interval of my awareness. But if this is the case, it is impossible for my state of awareness to span the interval of the object’s movement. At t2 I cannot be aware of the object moving from L1 to L10, because the object hasn’t got to L10 until t10 (ibid., p. 98). So my state of awareness of the object’s motion from L1 to L10 cannot span the interval t1–t10. This appears to contradict (1).
Soteriou proposes that the way to solve this puzzle is to adopt a relationalist theory of perception; and that once we do so we can reject (2). For relationalists, ‘there are phenomenally conscious states whose obtaining requires the obtaining of a relation of perceptual acquaintance, but which cannot be specified independently of that relation’ (ibid., p. 99). In the case we are currently considering, by (1) I am aware of the object’s motion from L1 at t1 to L10 at t10; this is my phenomenally conscious state. This phenomenally conscious state requires the obtaining of a relation of perceptual acquaintance, so the relationalist will say that the object’s motion from L1 at t1 to L10 at t10 is an event I stand in that relation to. This relation will hold over the period during which the object moves — viz. from t1 to t10. Hence, we can say that at any moment during that period I am in a state of awareness of the object’s motion from L1 to L10 — provided ‘we understand the claim as asserting that there are instants of time that fall within an interval of time over which [I am] aware of the movement of the object from L1 to L10’ (ibid., p. 106). In other words, although we can attribute these states of awareness, we can do so only in virtue of the fact that I am in a state of awareness over that period. As such, it is (at least in some cases) wrong to claim, as (2) does, that a subject’s mental state over a time is determined by their mental state at a time. Punctivism must be rejected.
3.2. Response to Soteriou
The relationalist may well be able to solve the puzzle by rejecting (2). This will not be challenged. Instead, it will be shown that, if a punctivist framework is accepted, (1) and (2) cannot together generate the puzzle. Contrary to Soteriou’s argument, (1) and (2) are troubling only for the holist.
We begin with the justification of (1). (1) is, Soteriou says, motivated by the transparency of experience. But this is an extremely controversial claim, and one which results in a severe dialectical handicap.
The transparency claim says that introspection only reveals properties of what an experience is of. This, however, is often taken as a warning against making the sort of inference Soteriou makes to (1). After all, to paraphrase Moore, I should not infer from the fact that I am having an experience of the colour blue that the experience itself is blue. Soteriou even acknowledges this point elsewhere, saying that one ‘might think Moore’s warning unnecessary, for there is little temptation to suppose that our mental episodes or states are actually blue in colour’ (2013, p. 12). But if transparency with respect to colour means I cannot infer that my experience of blue is itself blue, then, if experience is temporally transparent, surely I cannot infer from the fact that I am having an experience of an ordered succession that my experience itself is so ordered (see e.g. Tye, 2003, p. 96). And so, as Hoerl notes (2018, p. 134), ‘Soteriou uses the notion of transparency in a somewhat idiosyncratic fashion when it comes to talking about temporal features of experience’.
Now, it is not our intention to show that this idiosyncratic use is wrong (for a defence see Phillips, 2013; 2014a,b). Not only would it be beyond the scope of this paper, but a punctivist would be better off not committed either way. The point is that, by making such a controversial claim about temporal transparency, Soteriou’s solution will alienate many theorists. Indeed, since retentionalists hold that temporally extended events are represented by experiences which do not themselves so extend, their theory explicitly denies that the temporal features of an experience can be ‘read off’ from the phenomenology. Hence, Soteriou’s justification of (1) assumes the falsity of retentionalism. This weakens Soteriou’s argument on two fronts. Firstly, many punctivists are retentionalists. Soteriou’s argument will therefore do nothing to persuade these punctivists that they’re mistaken. Secondly, as discussed in Section 2, some retentionalists are holists (e.g. Lee), meaning that Soteriou’s argument will encounter resistance even within the holist camp.
In Soteriou’s defence one might reply as follows. Yes, his solution demands we accept a controversial understanding of temporal transparency, and yes, it is incompatible with retentionalism. But given that we are unable to make sense of motion perception without it, our only recourse is to rule out retentionalism as a viable model of temporal experience and embrace his controversial transparency claim.
The problem with this response is that, as we shall now see, Soteriou’s solution isn’t the only way to make sense of motion perception. In fact, it is only by implicitly rejecting punctivism that Soteriou is able to present his puzzle. And so, rather than speaking in favour of holism, his argument undermines it. It shows that holists cannot make sense of motion perception unless they commit themselves to relationalism, against retentionalism, and embrace a controversial understanding of transparency. The punctivist, meanwhile, is left wondering what all the fuss is about.
The reason for this can be brought out by considering Soteriou’s claim that the subject’s awareness of the object’s movement is a state of awareness. Without this claim, (1) and (2) won’t generate the puzzle. (2) is a claim about states, so the puzzle requires the subject’s awareness described in (1) to be a state of awareness in order to gain any traction. Our response is as follows: if the punctivist accepts that the proffered mental state obtains they will have to deny (2); and if they accept (2) they will have to deny that the mental state obtains. Either way, there is no puzzle for the punctivist.
Let’s first suppose the punctivist is willing to say that the subject is in the proffered state from t1 to t10. They will hold that:
(a) From t1 to t10 the subject is in a state of awareness of the object’s motion (viz. from L1 at t1 to L10 at t10).
Since they think that instantaneous mental states are fundamental, punctivists will give a reductive account of (a). They will say that what makes (a) true is that at t1 the subject is aware of the object at location L1, at t2 the subject is aware of the object at subsequent location L2, at t3 the subject is aware of the object at L3, and so on. And, in providing this sort of reductive account, they will be set against (2). Recall that (2) claims that if a subject is in a given state over a period of time, they will be in that same state at each moment during that period. As we can see, this is what the punctivist will deny in providing their reductive account of (a). As such, if the punctivist accepts the claim that the subject is in a state of awareness throughout the interval, they won’t accept (2). The puzzle doesn’t gain traction.
This position might encounter resistance. The objection would be that if a reductive account of (a) is true, then the awareness described in (1) would no longer belong to the category of ‘state’. What determines the category of a mental item ‘is the way in which that item fills the relevant period of time — whether it persists through time, or occurs during the time, or obtains throughout the time etc.’ (Steward, 1997, p. 73). If a period of time is filled by a mental item that can be given the sort of reductive account we have just seen, then that mental item cannot be said to obtain over time but rather to unfold over time. However, mental items that unfold don’t belong to the category of ‘state’; states obtain. So the mental item described in (a) would therefore be a state by name only.
The punctivist might then instead reject (a) and deny that the psychological property we attribute to the subject is a state. But if this is their position, (2) becomes irrelevant. (2) is a claim about states, and so it won’t be saying anything about the property we are attributing to the subject. Again, the puzzle fails to get off the ground. So whether or not the punctivist calls the mental item a state, they won’t be troubled by Soteriou’s puzzle. The puzzle can only be puzzling for the holist, who, if Soteriou is right, must then make several controversial commitments in order to get out of the tight spot they find themselves in.
3.3. Phillips’ argument
Phillips makes his case for holism in response to a puzzle raised by Delia Fara (2001), which results from the two following facts.
(3) We experience constant motion.
(4) Some positional changes are too small to be perceived.
(3) is obvious enough. An experience of constant motion is one during which the perceived object never appears stationary — such as when we watch a second hand sweep across a clock face. Similarly, there’s no denying (4) — take the motion of an hour hand over a few seconds as an example. However, by (3) there is a threshold positional difference below which we are unable to perceive change. Over small enough intervals, a constantly moving object won’t achieve this threshold and so won’t be seen to move. So an experience of motion will consist of many sub-intervals during which the object appears stationary. But an experience of constant motion is one in which the perceived object never appears stationary. (4) precludes (3).
Phillips’ solution requires we accept the following two claims and, in doing so, adopt the holistic conception:
Claim 1: Experience has a temporal field.
Claim 2: The content of the temporal field is purely determinable.
How these claims come together to solve Fara’s puzzle will emerge in our discussion of them. We begin with Claim 1. That experience has a temporal field is, Phillips claims, demonstrated by the fact that we are able to perceive the movement of some objects but not others. We can see the second hand sweep across the clock face, but the hour hand never appears to move. We only ever notice that it has moved. Our awareness is inferential, or indirect:
What this shows is that there is an upper bound to the stretches of time over which we can directly apprehend complete events and processes… we can think of our experience as having a ‘temporal field’, and of there being limits to the extent of the ﬁeld, say 300ms. (Phillips, 2011, p. 817)
The thought here is that we perceive the world through a small temporal window of around 300 ms, such that only if the positions of an object are sufficiently different to discriminate between within those 300 ms will the object’s motion be ‘genuinely an object of experience’ (ibid., p. 813). So either the positions are discriminable, in which case motion is represented over the 300 ms, or else the positions are indiscriminable, in which case no motion is represented at all.
Claim 2 says that the content of the temporal field is purely determinable. We can get an idea of determinable content by considering an analogous case. Consider the top edge of a sheet of paper. We know that, although the edge appears straight to the naked eye, a powerful enough microscope would reveal it to be rough and irregular. So we can say that the edge is at least ‘straightish’. ‘Straightish’ is determinable: a straightish line has many possible determinates — determinates which will differ from one another with respect to their micro-irregularities. Because we are unable to perceive these micro-irregularities, our experience represents the determinable ‘straightishness’ of the edge without representing its more determinate form. Thus, we can say that at such scales the content of our visuo-spatial field is purely determinable.
Phillips makes a similar claim about the content of the temporal field. We experience the second hand’s motion within the 300 ms window. But due to our perceptual limitations we are unable to determine any finer-grained details of the motion. There might be a change in velocity, for example, but such details are imperceptible. Thus, the content of sub-intervals of the temporal field is determinable, but not determinate.
This provides a solution to the puzzle. It follows from (4) that it is impossible to perceive continuous motion. But we can now distinguish between continuous and constant motion (as Phillips uses the terms). To perceive continuous motion it is necessary to perceive all the fine-grained details. But to perceive constant motion it is sufficient that the object never appears stationary. By Claim 2, the fact that we cannot perceive continuous motion does not entail that we cannot perceive constant motion. Provided the temporal content of the experience is determinable, we will perceive constant but not continuous motion. (4) does not preclude (3).
Only the holist can avail themselves of this solution. Since the content of the temporal field is purely determinable, this means that we cannot say anything about what is experienced during that interval other than what is experienced over the whole interval. Although it is true to say that we are perceiving motion at an instant, this is true only in virtue of the fact that that instant falls within the temporal field over which we are perceiving motion. What a subject experiences at instants within the temporal field is determined by what is experienced over the whole period. The extended experience of the temporal field is metaphysically prior to the instantaneous experiences it encapsulates. Punctivism must be rejected.
3.4. Response to Phillips
Our response to Phillips is as follows. His solution only supports holism on the assumption that retentionalism as traditionally understood is false. But when we follow Phillips in adopting an extensionalist framework, his notion of the temporal field turns out to have neurobiologically impossible consequences, and must be rejected.
By combining Claims 1 and 2, we get the conclusion that ‘[e]xperiential contents are fundamentally temporally extended: they are the contents of temporal ﬁelds’ (Phillips, 2011, p. 823). Let’s unpack this quote. It says, firstly, that experiential contents are fundamentally extended; and secondly, that these are the contents of temporal field. The first part remains silent as to whether the experiential event itself extends through time; it is only a claim about what gets represented. This is where Phillips’ notion of the temporal field comes in. The temporal field is naturally interpreted as a sort of extended experiential event. And so, if experiential contents are fundamentally extended, and if those contents unfold over time within extended experiential events, then the experiential events themselves will be fundamentally extended. This would then secure holism.
However, as Phillips acknowledges, not all theorists will accept this interpretation of the temporal field. The ‘temporal field’ is the name Phillips gives to what is sometimes called the ‘specious present’. And whilst extensionalists construe this as an extended experiential event, retentionalists construe it as an extended content — a content which, as discussed, is traditionally attributed to an instantaneous experience. As such, the traditional retentionalist seeking to avail themselves of Phillips’ solution will accept that experiential contents are fundamentally extended; but they won’t be compelled to accept that experiences themselves are. Hence Phillips’ argument should not be taken as a stand-alone argument against punctivism; it is an argument that supports holism once we have already rejected the retentionalist framework. True, Phillips argues against retentionalism elsewhere (e.g. 2014b), and assumes an extensionalist framework in his solution to Fara’s puzzle. But be that as it may, retentionalists unmoved by such challenges will have no reason to accept holism.
Even when we assume an extensionalist framework, however, Phillips’ solution runs into difficulties. Phillips’ idea of the temporal field is that of a 300 ms-or-so window through which ‘we can directly apprehend complete events and processes’. This is an appealing idea, and one that seems to fit with the phenomenology. The problem is that once we entertain that thought, we slide uncritically to the idea that we perceive motion over the duration of that interval; and this is what provides the platform for the claim that what we perceive at each part of that interval is determined by what we perceive over the interval as a whole.
We must, however, resist the idea that we perceive motion over the duration of the temporal field. The human eye has an angular resolution limit of somewhere between 0.17 and 0.47 arcminutes. Because of this, there is a minimum distance a stimulus must move before it is registered as being in a different location. We must therefore allow for an initial period before the object achieves this distance (which will vary depending on the object’s speed and distance from the observer). Once it does achieve this distance, we must then also allow for the time taken for neural processing. It takes 30–60 ms for the photoreceptors in the retina to process the signal (depending on retinal location), then ~50 ms to reach the brain. We therefore expect the signal to reach the brain about 100 ms later (for discussion of signal processing in monkeys, see e.g. Maunsell and Gibson, 1992). Further, the idea of ‘conscious perception’ (i.e. our ability to be explicitly aware of what we are seeing) is a highly complex neural process that evidence from neuroimaging and neurostimulation studies suggests requires a combined bottom-up (sensory) and top-down (predictive) approach in order to function effectively (see Pascual-Leone and Walsh, 2001; Dux et al., 2010; Meyer, 2012; Panichello, Cheung and Bar, 2013). Indeed, although the physiological correlates of ‘conscious perception’ are still largely unknown, it is widely accepted that visual signals need to traverse a diverse network of cortical areas in order to reach the level of explicit awareness. So we must first allow the time taken for an object to move a sufficient distance to be picked up by the eye; we must then allow the time taken for the signal to reach the brain; and we must then allow the time taken for the appropriate processing in the brain to take place before we see the object as moving. It should therefore be uncontroversial that within, say, 1 ms of observing a moving object, it is neurobiologically impossible to perceive its motion.
Consider what we might call a borderline motion experience. This is the experience (Phillips thinks) we would have when an observed object moves between two minimally discriminable positions just within the timeframe of the temporal field (whether that is 300 ms or some other interval — this is not important for the objection). According to Phillips’ notion of the temporal field, since the object’s positions are discriminable within this time frame, the positional change will be an object of experience. But, because the positions are minimally discriminable, if the object were to have moved any less in that time frame, its positions would be indiscriminable and so (according to Phillips) it would not be seen to move.
If it is impossible to perceive motion immediately upon perceiving an object’s position, then we must allow for some initial period after the first position is registered, but during which motion is not experienced. But this isn’t how the temporal field works for Phillips. According to Phillips, we ‘directly apprehend’ the ‘complete event’ of the object’s discriminable positional change. As the positional change is minimally discriminable within the temporal field, this means we should perceive motion over the duration of the temporal field. But as we saw, that’s impossible; the motion phenomenology will begin some way into the temporal field. Hence, motion will be perceived between indiscriminable positions over a sub-interval of the temporal field. Something’s not right with the idea behind Claim 1.
This also exposes a shortcoming with Claim 2 — that the content of the temporal field is determinable. The content of the temporal field could only be claimed to be determinable if it could first be claimed that motion is perceived over its duration. After all, to say that the content of the temporal field is purely determinable is to say that the most detailed description we can give of it is a description of its content over its entirety. If motion phenomenology begins some way into the temporal field, then we can describe the temporal field as comprising sub-intervals of differing contents. The content of the temporal field can’t therefore be purely determinable. So Claim 2 isn’t right either.
In Phillips’ defence, one might be tempted to reply along the following lines. We have claimed that it is impossible to experience motion immediately upon observing a moving object. But a holist like Phillips could insist that the question of what is experienced right away has no answer independent of what is experienced over the interval as a whole. It would be like asking whether a person who had just started to move was walking or running; it depends on what happens next.
This reply won’t stick, though. Firstly, the analogy doesn’t quite hold up. Both parties (we hope) agree that motion experience has neural correlates. And, as argued, these correlates will be absent for an initial period. So if the holist wants to insist that it is even derivatively true that motion is experienced right away, they will be admitting that, on their view, we can truthfully say that a subject can experience motion when those correlates are absent. They will be admitting that, so long as those neural correlates are present in the future, we can truthfully say of the experience something which, neurobiologically speaking, cannot be true of it.
Secondly, the reply sounds suspiciously circular. The justification Phillips gives us for the holistic conception is provided by Claim 2: since the content of the temporal field is determinable, truths about contents-at-instants depend on truths about contents-over-intervals. So it would have to first be established that the content is determinable in order to then claim that there is no non-derivative answer to the question of what is immediately experienced. And, as discussed above, the content of the temporal field could only be claimed to be determinable if it could first be claimed that motion is perceived over its duration. But that’s precisely the point we are challenging! Even if we allow, for the sake of argument, that experiences can be fundamentally extended, we are still waiting for an explanation of how the extended experience can have the extension Phillips is claiming it to have: how can it be true to say that the experience of motion begins immediately after first observing the object? No explanation for this is given, other than what is asserted in Claim 1: that the movement is directly perceptible within the bounds of the temporal field. So in order to lend plausibility to Claim 1, we look to Claim 2; but the latter can’t save the former unless the former is assumed to be true.
Admittedly, when we first look at a moving object there does not seem to be an initial period of stasis. When we are about to cross the road and look up at a moving car, the car does not appear stationary before suddenly jolting into motion. This is quite true. But it does not follow that during the initial period we must be experiencing it as moving. The car will only appear stationary if it is represented in experience as stationary. It does not appear as stationary, so clearly it is not represented as stationary. But it does not follow that during that period it must therefore be represented as moving. The details of the car’s motion could simply be omitted from the representation. So, during the initial period, although we do not experience the car as moving, we don’t experience it as not moving either. Our experience of the car will be indeterminate in this regard (that is to say, there will be indeterminacy as to what is represented, not that the representation will be a representation of indeterminacy).
Perhaps there are modifications that a staunch defender of Phillips’ proposal might want to try. At this point, however, we should begin to wonder whether there isn’t a more straightforward solution to Fara’s puzzle available. That is precisely what we now want to propose.
4.1. A punctivist proposal
We should reject Phillips’ notion of the temporal field as a sort of perceptual ‘window’ within which motion events are the objects of our experience. There will be an initial period upon observing a moving object during which motion is not represented. The motion phenomenology then ‘kicks in’ when a sufficiently great positional difference is observed in a sufficiently short time frame, and after a small amount of time is allowed for the corresponding neural processing. This presents a much simpler solution to Fara’s puzzle, as captured in the following principle of motion perception (PMP):
PMP: A subject enjoys (non-illusory) motion phenomenology when and only when an observed object is (subpersonally) registered in a sufficiently different position from a sufficiently recent position it was (subpersonally) registered in, and sufficient time has passed for the appropriate neural processing to occur.
Once we equip ourselves with PMP, a straightforward solution to the puzzle presents itself. To illustrate, let’s say that we begin observing a constantly moving object at t0. At t10, the object has achieved the sufficient positional difference within the necessary time frame, and the neural processing has taken place. At t10 we would begin to perceive motion:
At t11, the position of the object will be indiscriminable from its position at t10. This, according to Fara’s puzzle, means that we will not perceive motion. This is wrong, though. It is wrong because, although at t11 the position observed is insufficiently different from the position observed at t10, it is sufficiently different from its observed position at t1:
And so we would perceive motion from t10 onwards. But when, one might ask, is t10? Consider again a borderline motion experience. If the object’s positions are minimally discriminable within the 300 ms time frame, doesn’t our model imply that we will be looking at the object for 300 ms before we start to experience motion? One might want to object that this is too long a delay.
In fact, our model need make no such commitment. 300 ms is not our figure, and we reject Phillips’ notion of a temporal field. If it turned out that motion was experienced before 300 ms, then that would be the figure we’re looking for. Nevertheless, we are considering a borderline motion experience, and, as stipulated, the positions of the object only become discriminable at either end of the 300 ms (or whatever the figure may be). So how is it possible to perceive motion between positions that are indiscriminable? Rejecting the notion of the temporal field gives us (at least) two options. On the one hand, we could deny that conscious positional discrimination is necessary for motion perception (i.e. we could claim that it is possible to perceive motion between indiscriminable positions); on the other hand, we could allow that positional discrimination and motion perception occur at the same time, but that this happens sometime before 300 ms.
There may be empirical reason to doubt the latter option. Evidence suggests that everyday instances of motion perception might comprise (at least) two distinct perceptual mechanisms — one for positional change and one for what we might call ‘pure’ motion (construed as a sort of ‘visual sensation’ of motion). In the ‘waterfall illusion’, a scene seems to be moving despite its visual features remaining within one’s fixed visual field. One plausible interpretation is that the illusion occurs because a mechanism that results in ‘pure motion’ phenomenology is active but the mechanism for positional change is not (e.g. Mellor, 1988; Le Poidevin, 2007). The fine-grain motion illusion lends support to this idea. In this experiment, two dots are flashed on a screen in quick succession. These dots are so close together that, when the time between flashes is short enough, the two dots appear as a single unmoving dot. Similarly, when the time between flashes is long enough to notice, the second dot appears to be in the same place as the first. The positions are, at the conscious level, indiscriminable. But when the time between flashes is around 50 ms — between the two extremes — the subjects report seeing a single moving dot. This again suggests that motion can be perceived in the absence of perceptible positional change. Further credence to this dual-mechanism proposal is given by akinetopsia, a condition that renders the subject unable to perceive motion, despite an awareness of positional change. Subjects report seeing the world as a succession of ‘stills’ (see e.g. Zeki, 1991). In contrast to the previous illusions, this appears to be an instance where only the mechanism for positional change is active.
If there are indeed two distinct perceptual mechanisms involved in everyday instances of motion perception, we can unpack PMP as:
PMP1: A subject can consciously discriminate between an object’s positions when and only when the observed object is (subpersonally) registered in a sufficiently different position from a previous position it was (subpersonally) registered in, and sufficient time has passed for the appropriate neural processing to occur.
PMP2: A subject enjoys (non-illusory) ‘pure’ motion phenomenology when and only when an observed object is (subpersonally) registered in a sufficiently different position from a sufficiently recent position it was (subpersonally) registered in, and sufficient time has passed for the appropriate neural processing to occur.
There is a prima facie implausibility in supposing that both perceptual mechanisms would be equally as sensitive. Of any two neural mechanisms that process sensory information, it is more plausible to suppose that one will be more sensitive (quicker to respond, etc.) than the other. If, as Phillips thinks, we would perceive the motion of a moving object before its positions become discriminable, the natural conclusion would be that the mechanism for pure motion is (in everyday instances) the more sensitive of the two. The fine-grain motion illusion would appear to support this, given that the subject is able to perceive motion between indiscriminable positions. Whatever the neurobiological facts turn out to be, our punctivist proposal is flexible enough to accommodate them.
4.2. Anticipated objections
It might be objected that our model assumes what Daniel Dennett (1991) calls ‘Cartesian materialism’, which erroneously conceives of experience like a movie, with us — the subjects — as the audience. On this view, sense-data are sent to the brain for processing before reaching the ‘finish line’, whereupon they are transduced into consciousness and presented to the subject. The ‘finish line’ need not be taken too literally. Although a naïve Cartesian materialist might think there is a specific location for the neural–phenomenal transduction, the finish line could instead be thought of by the Cartesian materialist as the culmination of neural processing of a special sort. The relevant point is that on this view there is a determinate matter of fact about what information has been transduced, and is therefore experienced by the subject, at any particular moment. By assuming Cartesian materialism we imagine that we could, if you like, stop the film reel, and inspect the frame being shown at that moment. We can always ask ‘at what moment did that particular content become conscious?’. But this question, the argument goes, betrays a misunderstanding:
…being an item in consciousness is not at all like being on television; it is, rather, a species of mental fame. Almost literally. Consciousness is cerebral celebrity — nothing more and nothing less. Those contents are conscious that persevere, that monopolize resources long enough to achieve certain typical and ‘symptomatic’ effects — on memory, on the control of behavior and so forth. (Dennett, 1993, p. 929)
There are two related criticisms that can be brought out here. Firstly, just as it makes no sense to ask at which moment a celebrity acquires fame, it makes no sense to ask at which moment a particular content becomes conscious. And yet, in saying that that motion is experienced at t10, our model apparently provides an answer to this question. Secondly, our model could be accused of implicitly assuming Cartesian materialism in that it assumes there is a definitive fact of the matter about what is experienced at an instant. When we reflect on our experience, we imagine there is some canonical version of the phenomenal events to which we have just been subject. And by constructing this narrative it seems plausible that there is a fact of the matter about what was being experienced at any particular instant. But, the objection goes, this narrative — and therefore the narrative-at-an-instant — is a fiction; there is no canonical version of the phenomenal events. There are at any one time many different versions, all competing for influence (fame) throughout the brain.
The first objection shouldn’t be much trouble. In our depiction of the punctivist model, the subject’s experience of motion determinately begins at t10. The objection here would be that the contents of consciousness aren’t the sorts of things that have determinate start points. Neural influence isn’t an on/off affair: it spreads, and is a matter of degrees. Thus, there can be no moment when the experience of motion becomes determinately conscious. This might be true. But if it is true, we can build it into our model. Instead of positing an on/off experience of motion, it can chart the increase of neural influence with time. Where our current model shows the determinate beginning of the motion experience, we can take this as the point at which the neural influence is sufficient to prompt a verbal report:
With respect to the first criticism, then, the punctivist need not assume a Cartesian materialist conception of consciousness. And, if we take the simple model as modelling neural propagation (which is, according to Dennett and others, all there is to consciousness) rather than the broadcastings of the Cartesian theatre, we have a way to respond to the second objection. The second objection was that the idea of a canonical narrative-at-an-instant is a fiction, because there are many different competing versions at any one time. Well, if we are now modelling the propagation of information from specific modules, there will be a ‘canonical version’ at an instant. Not of the stream of consciousness per se — because there is no such thing as a ‘canonical version’ (or so we are assuming for the sake of this objection). But there will be a single, determinate matter of fact about what sort of effects the activities of a particular module of the brain will be having on other parts of the brain; about how far the information has spread and its efficacy in terms of verbal reports, behaviour, and so forth. And, moreover, there will be a fact of the matter about these effects at an instant. This can be modelled by the punctivist.
Neither Soteriou nor Phillips succeeded in showing that we need an holistic framework in order to make sense of motion perception. The puzzle Soteriou raised could only be puzzling for the holist, who was then forced make several controversial commitments the punctivist is free from. Meanwhile, Phillips’ holistic solution to Fara’s puzzle relied on a troublesome notion of a ‘temporal field’ and had neurobiologically dubious consequences. The punctivist model we have proposed is simple, flexible, and accounts for our perceptual limitations. It is the punctivist, rather than the holist, who is best able to make sense of motion perception.
Chuard, P. (2011) Temporal experiences and their parts, Philosophers Imprint, 11 (11), pp. 1–28.
Dainton, B. (2013) The perception of time, in Dyke, H. & Bardon, A. (eds.) A Companion to the Philosophy of Time, Chichester: Wiley-Blackwell.
Dainton, B. (2017) Temporal consciousness, in Zalta, E.N. (ed.) The Stanford Encyclopedia of Philosophy, [Online], https://plato.stanford.edu/entries/ consciousness-temporal/.
Deering, M.F. (1998) The limits of human vision, 2nd International Immersive Projection Technology Workshop, Ames, IA: Sun Microsystems.
Dennett, D. (1991) Consciousness Explained, Boston, MA: Little, Brown & Co.
Dennett, D. (1993) The message is: There is no medium, Philosophy and Phenomenological Research, 53 (4), pp. 919–931.
Dux, P.E., Visser, T.A.W., Goodhew, S.C. & Lipp, O.V. (2010) Delayed reentrant processing impairs visual awareness: An object substitution masking study, Psychological Science, 21, pp. 1242–1247.
Fara, D.G. (originally D. Graﬀ) (2001) Phenomenal continua and the sorites, Mind, 110, pp. 905–935.
Grush, R. (2007) Time and experience, in Müller, T. (ed.) Philosophie der Zeit, Frankfurt: Klosterman.
Hoerl, C. (2013) A succession of feelings, in and of itself, is not a feeling of succession, Mind, 122 (486), pp. 373–417.
Hoerl, C. (2017) Temporal experience and the philosophy of perception, in Phillips, I. (ed.) The Routledge Handbook of Philosophy of Perception, Oxon: Routledge.
Hoerl, C. (2018) Experience and time: Transparency and presence, Ergo, 5 (5), pp. 127–151.
Husserl, E. (1905/1964) The Phenomenology of Internal Time-Consciousness, Heidegger, M. (ed.), Churchill, J.S. (trans.), Bloomington, IN: Indiana University Press.
Kon, M. & Miller, K. (2015) Temporal experience: Models, methodology, and empirical evidence, Topoi, 34, pp. 201–216.
Le Poidevin, R. (2007) The Images of Time, Oxford: Oxford University Press.
Lee, G. (2009) Consciousness and the Passing of Time, doctoral dissertation, New York University.
Lee, G. (2014) Temporal experience and the temporal structure of experience, Philosophers’ Imprint, 4 (3).
Maunsell, J.H. & Gibson, J.R. (1992) Visual response latencies in striate cortex of the macaque monkey, Journal of Neurophysiology, 68 (4), pp. 1332–1344.
Mellor, D.H. (1988) Crane’s waterfall illusion, Analysis, 48 (3), pp. 147–150.
Meyer, K. (2012) Another remembered present, Science, 335, pp. 415–416.
Moore, G.E. (1903) The refutation of idealism, Mind, 12 (48), pp. 33–53.
Panichello, M.F., Cheung, O.S. & Bar, M. (2013) Predictive feedback and conscious visual experience, Frontiers in Psychology, 3, art. 620.
Pascual-Leone, A. & Walsh, V. (2001) Fast backprojections from the motion to the primary visual area necessary for visual awareness, Science, 292 (5516), pp. 510–512.
Phillips, I. (2011) Indiscriminability and experience of change, The Philosophical Quarterly, 61 (245), pp. 808–827.
Phillips, I. (2013) Perceiving the passing of time, Proceedings of the Aristotelian Society, 113 (3), pp. 225–252.
Phillips, I. (2014a) Experience of and in time, Philosophy Compass, 9 (2), pp. 131–144.
Phillips, I. (2014b) The temporal structure of experience, in Arstila, V. & Lloyd, D. (eds.) Subjective Time: The Philosophy, Psychology, and Neuroscience of Temporality, pp. 139–158, Cambridge, MA: MIT Press.
Prosser, S. (2016) Experiencing Time, Oxford: Oxford University Press.
Soteriou, M. (2007) Content and the stream of consciousness, Philosophical Perspectives, 21, pp. 543–568.
Soteriou, M. (2013) The Mind’s Construction, Oxford: Oxford University Press.
Steward, H. (1997) The Ontology of Mind: Events, Processes and States, Oxford: Oxford University Press.
Strasburger, H., Huber, J. & Rose, D. (2018) Ewald Hering’s (1899) On the Limits of Visual Acuity: A translation and commentary: With a supplement on Alfred Volkmann’s (1863) Physiological Investigations in the Field of Optics, i-Perception, 9 (3), 2041669518763675.
Throson, J., Lange, G.D. & Biederman-Thorson, M. (1969) Objective measure of the dynamics of a visual movement illusion, Science, 164, pp. 1087–1088.
Tye, M. (2003) Consciousness and Persons: Unity and Identity, Cambridge, MA: MIT Press.
Watzl, S. (2013) Silencing the experience of change, Philosophical Studies, 165, pp. 1009–1032.
Zeki, S. (1991) Cerebral akinetopsia (visual motion blindness): A review, Brain, 114, pp. 811–824.
Paper received June 2020; revised September 2020.
 For a valiant effort cataloguing the various positions, see Kon and Miller (2015).
 The lines of demarcation between these three camps can become blurred upon closer inspection. For instance, as Christoph Hoerl (2017) suggests, the difference between the snapshot and extensional models might appear to turn merely on whether or not one calls a conjunction of successive momentary experiences a single extended experience. We can, however, use the question of metaphysical priority to cut between the two. By positing instantaneous experiences, the snapshot model is incompatible with the holistic conception: if what we experience at an instant is determined by what we experience over an interval (as the holist claims), then our temporal phenomenology over that interval cannot be ‘built up’ from momentary snapshots during it (as the snapshot theorist claims). In positing instantaneous experiences, the snapshot theorist must therefore be a punctivist. Now, extensionalism could operate within a punctivist framework. The extensionalist holds that our experience of temporal phenomena unfolds in experiences that are themselves extended. But they could nevertheless maintain that these extended experiences have fundamentally instantaneous parts. If that was their view, they would then owe us an explanation as to what the difference between their position and the snapshot theorist’s really boiled down to. If they were to adopt the holistic conception, on the other hand, their position is clearly demarcated: not only are experiential events extended, they are fundamentally extended. That is not to say the extensionalist would be unable to demarcate their position unless they assign metaphysical priority to extended experiential events. Hoerl (2017) suggests the extensionalist could do so by aligning themselves with the relational view of perception. Some extensionalists might not be prepared to make such a commitment, however. Whether or not they are so prepared, they can be holists.
 For example, ‘retentional’ models are sometimes called ‘intentional’ (e.g. Hoerl, 2013) or ‘orthogonal’ (Dainton, 2013) models; ‘snapshot’ models are sometimes called ‘cinematic’ models (Dainton, 2017).
 There may be other reasons to question the holist’s walking example. For instance, as noted by a reviewer of this paper, the at-at theory of motion plausibly warrants a punctivist understanding of walking.
 Theorists typically distinguish between a positive and a negative version of the transparency claim. The positive claim says that when we try to attend to the features of our experience, we do so by attending to what the experience presents us with — what the experience is of. The negative claim says that these are the only features we can detect. It is this negative claim that Soteriou (2013, p. 88) invokes, and which we shall focus on.
 It should be noted, though, that Moore goes on to reject the transparency claim.
 That the temporal structure of an experience can be inferred from the phenomenology has faced objections in the form of empirical counter-examples (see Grush, 2007; Lee, 2009; 2014; Watzl, 2013). For responses to these counter-examples, see Phillips (2013; 2014a).
 Different punctivists will flesh this out in various ways. For example, a punctivist might hold that each instantaneous state is phenomenologically dynamic (Prosser, 2016) or non-dynamic (Chuard, 2011); or that they represent the object’s location over some small preceding interval (cf. Husserl, 1905/1964).
 The 300 ms is for illustrative purposes; as Phillips suggests, the figure would be expected to vary depending on the percept, subject, conditions, etc.
 We will follow Phillips in assuming an extensionalist framework for our criticism of his solution, but it should be noted that the positive case we go on to make for punctivism makes no such assumption.
 The eye cannot resolve isolated points below 0.47 arcminutes (Deering, 1998), but mismatches in alignment can be perceived down to around 0.17 arcminutes — which is counter-intuitively much smaller than the size of the retinal receptors (Strasburger, Huber and Rose, 2018).
 Our thanks to a reviewer for suggesting this reply.
 An alternative explanation might posit a delay-to-consciousness (for the motion processing to ‘catch up’ with positional detection) in conjunction with Husserlian ‘protentions’ to counteract the lag. See Grush (2007) for a model along these lines. The model would come under the ‘retentionalist’ banner, however, and so, following Phillips, it is discounted from the present discussion.
 See e.g. Throson, Lange and Biederman-Thorson (1969). A caveat: although the subjects have an experience as of motion, it is nevertheless an illusion. They are being presented with two stationary dots separated by a brief interval, rather than a single, continuously moving dot. Nevertheless, it demonstrates that perceptible positional change is not necessary for motion phenomenology.