Phenomenal Disunities in the Split Brain

1 Introduction

Starting in the 1960s, new forms of neuropsychological experiments were conducted on “split-brain patients” (henceforth “SB-patients”). SB-patients are individuals whose corpus callosum— the nerve tract connecting the cerebral hemispheres—has been surgically severed. The experiments (e.g., Gazza-niga et al., 1965; Gazzaniga & Sperry, 1967; Sperry, 1968, 1974; Sperry et al., 1969) revealed startling behavioural abnormalities. Under controlled conditions, patients’ hemispheres were incapable of transmitting information to each other.

It didn’t take long before SB-patients started to interest philosophers as well (Bayne, 2008, 2010; Davis, 1997; Hurley, 1998; Lockwood, 1989; Marks, 1980; Nagel, 1971; Puccetti, 1981; Schechter, 2018; Tye, 2003). Such patients seem to present cases where normal standards for attriuting concepts such as “mind,” “personhood,” “agency,” and “consciousness” break down. As Nagel wrote in one of the first philosophical papers published on SB-patients,

The feature of the mentalist conception of persons which may be recalcitrant to integration with these data is not a trivial or peripheral one, that might easily be abandoned. It is the idea of a single person, a single subject of experience and action, that is in difficulties. (Nagel, 1971: 396.)

This paper will be concerned with implications of SB-patients for the unity of consciousness. Recent years have seen defences of the view that split-brain consciousness is phenomenally unified despite being access disunified (Alter, 2010; Bayne, 2008, 2010; Bayne & Chalmers, 2003). Roughly, this theory posits that SB-patients have a unified phenomenology but limited capacities for behaviour requiring interhemispheric coor-dination. In opposition to this theory, I’ll argue that some SB-patients have experiences that are both access disunified and phenomenally disunified (provided that the patients are single subjects of expe-rience). For these patients, it’s not true that there’s a single subject with a single stream of conscious-ness without phenomenally disunified elements.

§2 presents an argument for why some SB-patients have phenomenally disunified experiences. In §3, the notions of phenomenal unity and access unity are explained. Then, §4, I’ll defend the premises of the argument presented in §2. In §5, I’ll outline implications of the argument.

2 Arguing for Phenomenal Disunity

Experiments on split-brain patients (henceforth SB-experiments) attempt to determine to what extent surgically disconnected hemispheres can transmit and integrate information without reliance on the corpus callosum or behavioural intermediaries, such as movement of patients’ eyes and hands.

We’ll focus on SB-experiments testing visual perception (e.g., Gazzaniga et al., 1965; Sperry, 1974; Sperry et al., 1969). The SB-patient is seated in front of two screens, one to either side of a fixed point the patient is instructed to gaze at. These screens obscure the patient’s hands. Images are then flashed for 100 milliseconds on either screen. The short screening-time means the patient cannot move her eyes and expand the range of visual information available to each hemisphere.

In one experiment (Sperry, 1974: 9), a patient is presented with the word “key” in her left visual field (LVF) and with the word “ring” in her right visual field (RVF). When asked what she sees, the patient responds “ring” and denies seeing another word. However, when instructed to select an item corresponding to the presented word using her left hand, the patient selects a key, even though both a ring and a keyring are available. When asked to describe the item, the patient says she retrieved a ring.

The patient’s behaviour can be explained as follows. Visual perception is contralateralized: Information on visual stimuli in RVF is transmitted to the left hemisphere (LH) while information on visual stimuli in LVF is transmitted to the right hemisphere (RH). Because LH typically controls speech, and because the patient cannot interhemispherically integrate information using the corpus callosum or eye movements, the patient will only verbally report visual stimuli presented in RVF (i.e., “ring”). However, because hand motor control is also largely contralateralized, the left-handed identification task is executed by the RH. RH can only access information about visual stimuli in LVF. Thus, the patient retrieves a key rather than a ring or a keyring. Finally, the patient claims to have identified a ring because no new information has been transmitted to LH.

Let’s call the patient in the keyring experiment “S,” her experience of “ring” “e₁,” and her experience of “key” “e₂.” I’ll argue that, if S is a single subject of experience, e₁ and e₂ are phenomenally disunified:

P1: When S is simultaneously presented with “ring” in RVF and “key” in LVF, S has two (simultaneous) phenomenally conscious experiences e₁ and e₂, with e₁ corresponding to “ring” and e₂ corresponding to “key.”

P2: e₁ and e₂ are access disunified.

P3: If e₁ and e₂ are access disunified, e₁ and e₂ are phenomenally disunified.

——————————————————————————————————————

C: e₁ and e₂ are phenomenally disunified.

To understand this argument, the notions of phenomenal unity, access unity, and conscious disunity must be explained.

3 Explicatory Interlude: Unity

We’ll begin by distinguishing between vehicles and contents of experience. The vehicle of an experi-ence e is the spatiotemporal and causal occurrence that’s involved in a subject’s undergoing e. If one is a physicalist, one can identify vehicles of experience with neurophysiological events in the brain. In turn, the content of e is what e is an experience of—how the world is presented or represented for the subject undergoing e.

A stance won’t be taken here on how to analyze contents of consciousness. Perhaps they’re truth-conditional, propositional contents. But they could also be identified with (apparent) objects, qualities, and relations one is consciously presented with when undergoing experience.

Both unity relations described below will be characterized using the notion of experiential con-tents. However, we’ll have reason to discuss vehicles of experience in §4.3.1.

3.1 Phenomenal Unity

A mental state is phenomenally conscious iff there’s something it’s like for a subject to be in that state (Block, 1995: 230). Examples of phenomenally conscious mental states are sensory states such as seeing a tomato, smelling a rose, or hearing the sound of a trumpet.

Correspondingly, the notion of phenomenal unity captures how, when a subject S has two (simultaneous, or close-to-simultaneous) phenomenally conscious mental states e₁ and e₂, there’s something distinctive it’s like for S to be in e₁ and e₂ together. For example, suppose e₁ is a visual experience of alphabetical characters appearing in a Microsoft Word document and that e₂ is an auditory experience of typing. There’s something it’s like to be in e₁, and there’s something it’s like to be in e₂. But if e₁ and e₂ are phenomenally unified, there’s also something distinctive it’s like to be in e₁ and e₂ together: A distinctive phenomenology associated with undergoing both e₁ and e₂ (simultaneously or in close temporal succession).

Tim Bayne (2010) further develops this idea. Bayne understands phenomenal unity mereologically. On his view, phenomenally unified experiences are constituents of more expansive, subsuming ex-periences:

[…] two conscious states are phenomenally unified when, and only when, they are co-subsumed [---]. What it is for one experience to subsume another is for the former to contain the latter as a part. (Bayne, 2010: 20-21.)

Bayne’s account doesn’t specify the relata of mereological relations underlying phenomenal unity. Are the relata contents of experience or vehicles of experience? The right answer is contents. Phenomenal unity is a relation characterizing a subject’s phenomenology. It pertains to how contents are presented or represented in experience. We can thus modify Bayne’s characterization of phenomenal unity as follows:

Phenomenal Unity: Necessarily, for any two experiences e₁ and e₂ with contents c1 and c2 and subject S, e₁ and e₂ are phenomenally unified for S iff S, when undergoing e₁ and e₂, has an experience E with content c3, where c3 subsumes c1 and c2.

Let’s return to the example of a visual experience (e₁) of alphabetical characters and an auditory experience (e₂) of typing. If e₁ and e₂ are phenomenally unified, undergoing those experiences involves having a more expansive experience whose content includes both alphabetical characters and typing sounds. In contrast, if e₁ and e₂ aren’t phenomenally unified, there’s no experience with a subsuming content. S experiences alphabetical characters, and she experiences typing sounds, but she doesn’t experience them together, as aspects of a single experiential presentation or representation of the world.

3.2 Access Unity

Block writes that a mental state is access conscious iff…

[…] a representation of its content is (1) […] poised for use as a premise in reasoning, (2) poised for rational control of action, and (3) poised for rational control of speech. (Block, 1995: 231).

For example, my perceptual experience of the cup of tea standing on my desk is access conscious. I’m able to report the presence of the cup, both verbally and non-verbally. Additionally, my experience facilitates rational action and inference; reaching out to the cup to quench my thirst, inferring that someone prepared the tea, and so on. In short, a mental state is access conscious for a subject S iff its content is available to S for rational control of action, inference, and report (verbal or nonverbal).

Correspondingly, the notion of access unity captures how two access-conscious mental states can be jointly available for rational control of action, inference, and report. We’ll characterize access unity as follows:

Access Unity: Necessarily, for any two mental states e₁ and e₂ with contents c1 and c2 and subject S, e₁ and e₂ are access unified for S iff (representations of) c1 and c2 are jointly available to S for verbal or non-verbal report, inference, and rational control of action.¹

If e₁ and e₂ are access unified, S can access the sum, or conjunction, of c1 and c2 in her control of report, inference, and rational action. As an example, suppose that, upon entering her kitchen, S has an auditory experience (e₁) of buzzing flies (c1) and an olfactory experience (e₂) of a stench (c2). If e₁ and e₂ are access unified, S can report, not merely on the buzzing and the stench individually, but on their joint occurrence. Moreover, S can engage in inference and action that are rationalized in light of the sum or conjunction of c1 and c2; e.g., inferring that food items must have gone bad, and telephoning pest controllers.

3.3 Disunity

Consider the claim that, necessarily, a subject’s experiences are unified (“Unity” for short).²

Let’s say that two experiences are disunified iff they constitute counterexamples to Unity. What would it take for e₁ and e₂ to be disunified?

It would be inadequate to say that e₁ and e₂ are disunified iff they aren’t unified. For there are many non-unified experiences that don’t threaten Unity. My experiences and your experiences aren’t unified, nor are experiences I had as a ten-year old unified with experiences I’m having as an adult writing this paper. Since these experiences aren’t eligible candidates for standing in unity relations, they don’t constitute counterexamples to Unity.

So we need a more discerning characterization of disunity. It will be assumed that two experiences e₁ and e₂ are eligible candidates for phenomenal (or access) unity and phenomenal (or access) disunity iff four conditions are satisfied:

(i) e₁ and e₂ must have the same subject.

(ii) e₁ and e₂ must both be phenomenally (or access) conscious mental states.

(iii) e₁ and e₂ must be simultaneous.

(iv) e₁ and e₂ must lack phenomenal unity (or access unity).

The third stipulation is controversial. As far as access unity is concerned, it seems incorrect. Two contents can be jointly available for rational action, inference, and report even though they aren’t presented or represented as simultaneous. In the example above, S might experience the smell first and hear the buzzing sounds later.

However, applied to phenomenal unity, the simultaneity requirement is more prima facie compelling. For e₁ and e₂ to be eligible for phenomenal unity, their contents, c1 and c2, must be experienced together, as aspects of a subsuming content c3. And it may be thought that a precondition for c1 and c2 to be experienced as aspects of a subsuming content of experience is that e₁ and e₂ are simultaneous. This would mean that, in order to constitute a non-trivial counterexample to the phenomenal unity of consciousness, e₁ and e₂ would have to be simultaneous. I’ll question this picture in §4.3.1, but we can proceed assuming its correctness.

4 Back to the Argument: Defending the Premises

We’re now in a better position to understand the argument presented in §2. P1 states that S has two phenomenally conscious experiences in the keyring experiment, one (e₁) or “ring” and one (e₂) of “key.” P2 posits that these experiences are access disunified. Finally, P3 states that the access disunity of e₁ and e₂ entails their phenomenal disunity. While S has an experience with the content “ring,” and an experience with the content “key,” S doesn’t have an experience with the content “keyring” (unlike a neuroanatomically normal subject, who would have visually experienced “keyring”).

The argument’s conclusion should be interpreted as the conditional claim that, if e₁ and e₂ have the same subject, then they would be phenomenally disunified. If there were two subjects of experience associated with the patient, one corresponding to each hemisphere, e₁ and e₂ wouldn’t be phenomenally disunified. “S” is just shorthand, and the argument doesn’t show, nor presuppose, that there’s a single subject of experience corresponding to S. What the argument’s conclusion does rule out is the position that S is a single subject with a single stream of consciousness without phenome-nally disunified elements (see, e.g., Alter, 2010; Bayne, 2008, 2010; Bayne & Chalmers, 2003).

It will be useful to consider the premises in mixed order: First, P2; second, P3; and third, P1.

4.1 Access Disunity

P2 states that S’s experience of “ring” and her experience of “key” are access disunified. To show this, two things must be established: Firstly, that e₁ and e₂ both are access conscious, and secondly, that e₁ and e₂ aren’t access unified.

S’s behaviour in the experiment supports the hypothesis that both e₁ and e₂ are access conscious. When it comes to e₁, the content of the experience, “ring,” is available to S for verbal report. The case of e₂ is less straightforward. Given RH’s limited capacity for speech, S cannot verbally report that she saw “key.” However, S performs an advanced discrimination task, in which she identifies a key out of a range of alternatives. If “key” weren’t available to S for control of (rational) action, there wouldn’t be an explanation of S’s capacity to execute the discrimination task. Thus, both e₁ and e₂ are access-conscious.

This leaves us with the second point: Access disunity between e₁ and e₂. If e₁ and e₂ had been access unified, “ring” and “key” would have been jointly available to S for verbal or non-verbal report, inference, and rational control of action. On the supposition that “ring” and “key” are jointly available in this sense, two predictions follow about S’s behaviour. Firstly, it’s to be expected that S would have verbally reported, not only that she saw “ring” and “key,” but also the compound word, “keyring.” Secondly, it’s to be expected that, when executing the identification task, S would have selected the keyring, rather than the key or the ring.

Neither prediction is borne out by S’s behaviour. Neither “key” nor “keyring” is verbally reported by S. Likewise, S retrieves the key while ignoring both the ring and the keyring. In both cases, the contents of e₁ and e₂ are available to S only individually and not jointly.

Perhaps access unity is a notion admitting of degrees. In that case, it would be incorrect to infer that e₁ and e₂ aren’t access unified because there’s a single lapse in the joint availability of “ring” and “key.” If S had been able to perform the reporting task (verbally reporting “keyring”) but not the identification task (selecting the keyring), or vice versa, it would have been defensible to claim that e₁ and e₂ are access unified, just less so than an analogous pair of experiences undergone by a neuroanatomically normal subject. However, S fails at both tasks, as well as at the task of naming the object she identifies with her left hand. The joint availability of “ring” and “key” isn’t incomplete: It’s non-existent. Thus, e₁ and e₂ aren’t access unified.

Since e₁ and e₂ are access-conscious but not access-unified, e₁ and e₂ are access-disunified.

4.2 Access Disunity Without Phenomenal Disunity?

P3 states that, if e₁ and e₂ are access disunified, they’re phenomenally disunified. This move has come under critique. Bayne and Chalmers claim that,

It is plausible that in split-brain cases, there is some sort of breakdown of access unity […]. But […] a breakdown of access unity does not entail a breakdown of phenomenal unity. So the possibility re-mains that split-brain subjects have a unified phenomenal field […]. (Bayne & Chalmers, 2003: 38-39.)

Similarly, Alter suggests that “[splitting] the corpus callosum might […] impede neural mechanisms that underlie access consciousness but not phenomenal consciousness.” (Alter, 2010: 25.)

Pace these philosophers, I’ll argue that phenomenal disunity between e₁ and e₂ is the best explanation of access disunity between e₁ and e₂.³

4.2.1 Phenomenal Consciousness Explains Access Consciousness

If a sensory or perceptual state e is access-conscious for a human subject S, the best explanation is that e is phenomenally conscious.⁴ If S can report the content c of e and use c in inference and rational control of action, the best explanation is that there’s something it’s like for S to experience c when undergoing e.

This isn’t to say that phenomenal consciousness is conceptually necessary for access consciousness. For example, consider zombies—microphysical and behavioural duplicates of human beings that lack phenomenal consciousness (Chalmers, 1996). Zombies have access-conscious states but no phenomenally conscious states. However, the (prima facie) conceptual possibility of zombies doesn’t undermine the claim that, at the actual world, phenomenal consciousness is the best explanation of access consciousness.

Something that could undermine that claim would be an empirical case of access consciousness without phenomenal consciousness. A potential candidate is blindsight. Blindsight is a condition in which subjects claim to have no awareness in parts of their visual field but nevertheless can “guess” answers to questions about stimuli in blind areas with a success-level significantly above chance. Questions can be about the location, direction, and movement of visual stimuli. Even when informed that they’ve answered correctly, blindsight patients deny having been aware of the stimuli, and claim they simply guessed. For an overview on blindsight, see (Weiskrantz, 2009).

Blindsight may seem like a case of access consciousness without phenomenal consciousness. One might think the patient’s success in identifying visual stimuli couldn’t be explained unless representations of those stimuli were available for report. And this at least looks like access consciousness. In contrast, provided that one trusts the patient’s reports of being unaware of stimuli in parts of her visual field, one must infer that her (seemingly) access conscious representations of blind visual stim-uli aren’t phenomenally conscious.

However, this assessment is too hasty, as Block points out (1995: 233). If a mental state e is access conscious, it’s content, c, must be available to the subject, not only for report, but also for rational control of action and inference. Blindsight patients don’t satisfy this condition. Block mentions an experiment reported by Marcel (1986), in which a thirsty blindsight patient is presented with a glass of water in a blind area of her visual field. When asked, the patient can specify the location of the presented object (which isn’t described as a glass of water). However, she doesn’t reach out to the water to quench her thirst. The content of the patient’s visual representation isn’t available to her for rational control of action. Consequently, the representation isn’t access conscious.

4.2.2 Interlude: Split-Brain Patients and Blindsight

The foregoing remarks about blindsight may be thought to undermine my earlier defence (in §4.1) of the claim that S’s experience of “key” (e₂) is access conscious. For couldn’t S be “blindsighted” with regard to her LVF? If that were true, S would have a representation e₂ of “key,” but this representation wouldn’t be access-conscious. There cannot be access disunity between e₁ and e₂ if e₁ and e₂ aren’t both access-conscious states. And if e₁ and e₂ aren’t access disunified, P3 cannot be used to infer that e₁ and e₂ are phenomenally disunified.⁵

The problem with interpreting S as suffering from blindsight is that her behaviour is nothing like that of a blindsight patient. Blindsight patients cannot engage in rational action and inference based on representations of visual stimuli in blind areas (remember the water example). In contrast, in the keyring experiment, “key” is available to S for the execution of a complex identification task. When asked to identify an object corresponding to “key,” S acts rationally, comprehending the semantic properties of “key,” inferring what tactile properties would be characteristic of an object belonging to the semantically relevant kind (i.e., a key), and identifying the key from a range of alternatives based on those tactile properties. This is a purposeful, demanding task that’s different in kind from the blindsight patient’s idle guesses.

In sum, S isn’t blindsighted in the keyring experiment. Both e₁ and e₂ are access-conscious.

4.2.3 Absence of Phenomenal Consciousness Explains Absence of Access Consciousness

Ceteris paribus, the best explanation of a sensory or perceptual state e’s being access conscious is that e also is phenomenally conscious (see above). An implication of this is that, if e isn’t access conscious, the default hypothesis should be that e isn’t phenomenally conscious either.

This isn’t to say there cannot be phenomenally conscious states that aren’t access conscious. Block (1995: 234) discusses the example of a subject (let’s call her S*) noticing the sound of a drill after having had a conversation requiring significant concentration. Before noticing the sound, S* isn’t access conscious of it; a representation of the sound isn’t available to S* for rational control of inference, action, and report. Nevertheless, it’s still conceivable that S* heard the sound all along and that there was something it was like for her to do so. That would be a case of phenomenal consciousness without access consciousness.

The lesson seems to be that access consciousness requires attention. S* doesn’t attend to the drill sounds because she’s focused on her conversation. But this doesn’t impugn the rule that, if a sensory or perceptual state e isn’t access conscious and the subject’s attention isn’t monopolized by another object or activity, the best explanation of e’s not being access conscious is e’s not being phenomenally conscious.

4.2.4 Phenomenal Disunity Explains Access Disunity

The explanatory relation between phenomenal unity and access unity mirrors that between phenomenal consciousness and access consciousness. Firstly, access unity is best accounted for by phenomenal unity. When a subject has access unified experiences, it’s usually because they’re phenomenally unified. Suppose a neuroanatomically normal subject, S*, volunteered to undergo the keyring experiment. It’s reasonable to assume that S*’s experiences of “ring” and “key” would be access unified; S* would have reported that she had seen “keyring,” and would have selected the keyring in the identification task. The most straightforward explanation of S*’s success is that her experiences of “ring” and “key” were phenomenally unified; that she had an experience with the content “keyring.”

Secondly, if the explanatory relation between phenomenal unity and access unity is like that between phenomenal consciousness and access consciousness, this rule should inform our interpretation of S in the “keyring” experiment. Thus, the most straightforward interpretation of the structure of S’s consciousness is that e₁ and e₂ are access disunified because they’re phenomenally disunified.

However, just as there are cases where mental states are phenomenally conscious but not access conscious, there could perhaps be cases where experiences are phenomenally unified despite being access disunified. That possibility cannot be ruled out a priori. What an objection to P3 needs to show is that there are positive reasons for holding that S constitutes one of those exception cases.

One line of thought is inspired by what was said above about the link between access consciousness and attention. A phenomenally conscious mental state can be access unconscious if the subject doesn’t attend to the object or content of the state. Likewise, it may be thought that two experiences can be phenomenally unified despite being access disunified if the subject doesn’t simultaneously attend to their contents. One could thus deny P3 by claiming that split-brain surgery upsets the capacity of simultaneously attending to experiential contents, while leaving phenomenal unity intact.

However, this objection to P3 presupposes a conception of the relation between phenomenal consciousness and attention that’s more demanding than that envisaged by my interpretation of Block’s drilling example. I interpreted the drill case as a situation where a subject has a phenomenally conscious mental state but doesn’t attend to its content, thereby preventing the state from being access conscious. But, crucially, the subject doesn’t attend to the content of her mental state because she’s engaged in an activity monopolizing her attention. If the subject hadn’t been focused on her conversation, her auditory representation would have been access conscious. In contrast, in the keyring experiment, the situation is different. For e₁ and e₂ don’t simply happen to be access disunified; there’s no way for the subject to get around the access disunity by redirecting her attention. The objection implies that e₁ and e₂ are phenomenally unified despite being access disunified because it’s impossible for S to simultaneously attend to “ring” and “key.” Thus, the right analogy isn’t the drill example, but a phenomenally conscious state whose content is impossible to attend to. It’s a highly questionable assumption that there can be features of phenomenal consciousness that are impossible for a subject to attend to.

In conclusion, the fact that S’s experiences of “ring” and “key” are access disunified is best explained by the hypothesis that the experiences are phenomenally disunified.

4.3 Two Phenomenally Conscious Experiences?

I’ve assumed so far that P1 is true. That is, I’ve assumed that S has two phenomenally conscious experiences in the keyring experiment; e₁ of “ring” and e₂ of “key.” Prima facie, this assumption seems compelling. However, there are objections to P1. Tim Bayne argues that, in the context of SB-experiments, a SB-patient’s cerebral hemispheres aren’t conscious simultaneously but sequentially:

[…] we should think of [phenomenal] consciousness in the split-brain as moving or switching from one hemisphere to another [---]. The patient might be conscious of the word ‘key’ (due to right hemisphere activity), or she might be conscious of the word ‘ring’ (due to left hemisphere activity), but she will not be conscious of both ‘key’ and ‘ring’ at once, even when the two words are simultaneously presented to her. (Bayne, 2010: 210.)

Bayne’s theory implies that e₁ and e₂ aren’t both phenomenally conscious experiences (although they may both be access conscious states). Since S doesn’t have two phenomenally conscious experiences, there’s no instance of phenomenal disunity in the keyring experiment. Let’s call Bayne’s theory the Switching View. If the Switching View is true, P1 is false.

4.3.1 Explicating the Switching View

The Switching View has been presented as implying that S only has one phenomenally conscious experience in the keyring experiment. Let’s call this an eliminationist form of the Switching View, as it “eliminates” one of S’s phenomenally conscious experiences.

However, there’s another way to understand Bayne’s position. Simultaneous flashing of “ring” and “key” on the screens in the keyring experiment might result in two non-simultaneous phenomenally conscious experiences, one of “ring” and one of “key.” Let’s call this an accommodationist version of the Switching View, as it “accommodates” both of S’s phenomenally conscious experiences.

Eliminationism and accommodationism give different explanations of why P1 is false. According to eliminationism, P1 is false because S only has one phenomenally conscious experience. In contrast, an accommodationist explanation of the falsity of P1 is premised on what was said in §3.3 about conditions for disunity. There, we said, provisionally, that if e₁ and e₂ aren’t simultaneous, they aren’t eligible candidates for phenomenal unity or phenomenal disunity. Let’s call this the “Temporal Eligibility Principle” (TEP). According to accommodationism, P1 is false because e₁ and e₂ aren’t simultaneous. And this non-simultaneity, in conjunction with TEP, implies that e₁ and e₂ aren’t eligible candidates for phenomenal disunity.

Is TEP true? Before answering that question, the notion of “simultaneity” must be disambiguated. On one disambiguation, e₁ and e₂ are simultaneous iff the vehicles of those experiences, v1 and v2, occur simultaneously (relative to some inertial frame of reference). On another disambiguation, e₁ and e₂ are simultaneous iff the contents of those experiences, c1 and c2, are experientially presented or represented as being simultaneous.

If TEP is combined with a vehicular construal of simultaneity, it follows that e₁ and e₂ are eligible for phenomenal unity or phenomenal disunity only if v1 and v2 are simultaneous. But there’s no reason to accept this. It’s likely that many phenomenally unified experiences have vehicles (e.g., brain states) occurring at slightly different times. Correspondingly, there isn’t any reason to hold that phenomenally disunified experiences must have simultaneous vehicles.

One could make a vehicular construal of TEP more plausible by substituting a temporal proximity requirement for the simultaneity requirement. Perhaps e₁ and e₂ are eligible candidates for phenomenal unity or phenomenal disunity only if v1 and v2 are either simultaneous or occurring in close temporal succession. However, S’s experiences of “ring” and “key” do occur in close temporal succession. So, the revised form of vehicular TEP doesn’t warrant inferring that e₁ and e₂ are ineligible candidates for phenomenal unity or disunity.

That leaves us with a content-oriented interpretation of TEP. On this interpretation, if c1 and c2 aren’t experientially presented or represented as simultaneous, then e₁ and e₂ aren’t eligible candidates for phenomenal unity or phenomenal disunity. Applied to the keyring experiment, content-TEP implies that phenomenal disunity between e₁ and e₂ would have required that “ring” and “key” were experientially presented or represented as occurring simultaneously. Since “ring” and “key” aren’t presented or represented in this way (let us suppose), e₁ and e₂ aren’t phenomenally disunified.

The problem with content-TEP is that it makes phenomenal disunity between e₁ and e₂ conceptually incoherent. Suppose content-TEP is true. Now imagine that, in the keyring experiment, “ring” and “key” are experientially presented or represented as being simultaneous. What would it be like to be S in this case? It’s insufficient for characterizing what it’s like to be S to separately specify what it’s like for her to be in e₁ and what it’s like for her to be in e₂. For this wouldn’t capture the fact that S experiences “ring” and “key” as occurring simultaneously, as belonging to the same “time-chunk” in her temporal phenomenology. Since there’s an aspect of S’s phenomenology that isn’t captured by separately specifying what it’s like for her to be in e₁ and what it’s like for her to be in e₂, there’s something distinctive it’s like for S when “ring” and “key” are experientially presented or represented as being simultaneous. Therefore, e₁ and e₂ are phenomenally unified. So, it follows that, if e₁ and e₂ satisfy the eligibility condition in content-TEP, they must be phenomenally unified. Phenomenal dis-unity has been rendered conceptually impossible.

Phenomenal disunity may well be metaphysically or nomologically impossible. Perhaps there’s an argument establishing that conclusion. But the key word here is “argument.” It would be unaccepta-ble to pronounce, by mere fiat of terminological stipulation, that there cannot be such a thing as phenomenal disunity. For this reason, content-TEP should be rejected.

Neither disambiguation of “simultaneity” furnishes us with a defensible version of TEP which establishes that there’s no phenomenal disunity in the keyring experiment. Therefore, accommodationism is untenable. I’ll proceed assuming that Bayne is an eliminationist.

4.3.2 Responding to the Switching View

The Switching View, in its eliminationist form, denies that S has two phenomenally conscious experiences. This denial is remarkable, for S fulfils behavioural conditions that ordinarily would warrant the attribution of two phenomenally conscious experiences. S has access-conscious mental states corresponding both to “ring” and “key” (see §4.1). Bayne must explain why S has two access-conscious states but only one phenomenally conscious state.

Bayne cannot do this, for two reasons. Firstly, a perceptual or sensory state’s being access conscious but not phenomenally conscious is more difficult to explain than that state’s being phenomenally conscious but not access conscious. In the latter case, one can give the explanation that access consciousness is more cognitively demanding than phenomenal consciousness. As we saw above, access consciousness requires a minimum-level of attention. Further, it requires capacities for verbal or non-verbal report, inference, and rational action that may not be necessary for phenomenal consciousness. In contrast, to explain a perceptual or sensory state’s being access conscious but not phenomenally conscious, an account must be given of how the state has a mental feature (access consciousness) requiring complex cognitive characteristics even though it lacks a mental feature (phe-nomenal consciousness) not requiring such characteristics. In the case of S, this is extremely difficult.

Secondly, denying the mutual phenomenal consciousness of e₁ and e₂, while maintaining their mutual access consciousness, contravenes a principle that is fundamental both in philosophical interpretation of cognitive psychology and in everyday life. This principle is that both attributions of phenomenal consciousness and withholdings of such attributions can, and should, be warranted in light of facts about subjects’ behaviour. Without this principle, there’s no obvious way to structure philo-sophical debates on consciousness. This holds particularly true in connection with SB-patients. Neuroanatomically normal subjects cannot adopt the first-person perspective of SB-patients. We are, as Nagel writes, “unable to project ourselves into their mental lives.” (Nagel, 1971: 410.) Without deferring to behavioural data about SB-patients, it’s difficult to see how one could argue systematically for or against views about the structure of such patients’ consciousness.

Thus, the supposition that e₁ and e₂ aren’t both phenomenally conscious is problematic. Since S has two access-conscious, perception-related states, it should be inferred that S has two phenomenally conscious experiences in the keyring experiment.

5 Implications and Conclusion

The argument I’ve presented has a conditional conclusion: If e₁ and e₂ have the same subject, then they’re phenomenally disunified. Thus, the conclusion rules out theories according to which SB-patients are single subjects of experience with a single stream of consciousness without phenomenally disunified elements (Alter, 2010; Bayne, 2008, 2010; Bayne & Chalmers, 2003). This is significant, for it leaves us with two theoretical contenders that both must make nonintuitive and startling claims about the structure of consciousness. Let me explain.

What theoretical verdict one makes about the structure of S’s consciousness as she undergoes the keyring experiment will hinge on a fundamental question about the relationship between phenomenal unity and streams of consciousness. This is the “Unity Question:”

The Unity Question: If e₁ and e₂ are phenomenally conscious experiences whose vehicles v1 and v2 are simultaneous or occurring in close temporal succession, do e₁ and e₂ belong to the same stream of consciousness E only if e₁ and e₂ are phenomenally unified?

Suppose we answer “yes” to the Unity Question. Then it follows that S has two streams of consciousness when undergoing the keyring experiment. Her experiences of “ring” and of “key” aren’t phenomenally unified. Since the experiences have vehicles that are simultaneous or occurring in close temporal succession, it follows that S’s experiences belong to different streams of consciousness.⁶

Interpreting S as having two streams of consciousness involves an abandonment of the commonsense view that any human being, or, at least, any human animal, has a single stream of consciousness.⁷

Let’s suppose, alternatively, that we answer “no” to the Unity Question. Then, it would be possible to hold that S has a single stream of consciousness some of whose simultaneous, or close-to-simultaneous, elements are phenomenally disunified. On this view, S’s experiences of “ring” and “key” are phenomenally disunified, but nevertheless belong to the same stream of consciousness.⁸

For this kind of view, the challenge is to articulate an account of the individuation of streams of consciousness that doesn’t appeal to the phenomenal unity relation. There must be some explanation of how two phenomenally conscious experiences that are simultaneous, or occurring in close temporal succession, can belong to the same stream of consciousness despite their phenomenal disunity. What grounds the fact that experiences belong to the same stream of consciousness, if it isn’t unity relations? Regardless of what answer one gives to that question, it will diverge from an intuitive way of understanding (synchronic) relations of co-membership in a stream of consciousness, namely in terms of phenomenal unity relations.

In conclusion, SB-patients provide an invaluable source of data to philosophers of mind. The unity of consciousness is a topic that traditionally has been discussed in a priori terms. SB-patients bring this perennial topic of contention into contact with empirical reality. The outcome is both illuminating and challenging. Not all cherished notions about the unity of consciousness will with-stand critical scrutiny.

References

Alter, T. (2010). A Defense of the Necessary Unity of Phenomenal Consciousness. Pacific Philosophical Quarterly, 91(1), 19–37.

Bayne, T. (2008). The Unity of Consciousness and the Split-Brain Syndrome. Journal of Philosophy, 105(6), 277–300.

Bayne, T. (2010). The Unity of Consciousness (pp. 1–207). Oxford University Press.

Bayne, T., & Chalmers, D. (2003). What Is the Unity of Consciousness? In A. Cleeremans (Ed.), The Unity of Consciousness (pp. 23–58). Oxford University Press.

Block, N. (1995). On a Confusion about a Function of Consciousness. Brain and Behavioral Sciences, 18(2), 227--247.

Chalmers, D. (1996). The Conscious Mind: In Search of a Fundamental Theory (Vol. 4, pp. 609–612). Oxford University Press.

Davis, L. H. (1997). Cerebral Hemispheres. Philosophical Studies, 87(2), 207–222.

Gazzaniga, M. S., Bogen, J. E., & Sperry, R. W. (1965). Observations on Visual Perception After Disconnexion of the Cerebral Hemispheres in Man. Brain, Vol 88(2), 221–236.

Gazzaniga, M. S., & Sperry, R. W. (1967). Language after Section of the Cerebral Commissures. Brain, 90, 131–148.

Hurley, S. L. (1998). Consciousness in Action. Harvard University Press.

Lockwood, M. (1989). Mind, Brain, and the Quantum: The Compound ‘I’. Blackwell.

Marcel, A. J. (1986). Consciousness and Processing: Choosing and Testing a Null Hypothesis. Behav-ioral and Brain Sciences, 9, 40–44.

Marks, C. E. (1980). Commissurotomy, Consciousness, and Unity of Mind. MIT Press.

Nagel, T. (1971). Brain Bisection and the Unity of Consciousness. Synthese, 22(3–4), 396–413.

Puccetti, R. (1981). The Case for Mental Duality: Evidence from Split-Brain Data and Other Considerations. Behavioral and Brain Sciences, 4(1), 93–123.

Schechter, E. (2012). The Switch Model of Split-Brain Consciousness. Philosophical Psychology, 25(2), 203–226.

Schechter, E. (2018). Self-Consciousness and Split Brains: The Minds’ I. Oxford University Press.

Snowdon, P. F. (2014). Persons, Animals, Ourselves. Oxford University Press.

Sperry, R. W. (1968). Hemisphere Deconnection and Unity in Conscious Awareness. American Psychologist, 23, 723–733.

Sperry, R. W. (1974). Lateral Specialization in the Surgically Separated Hemispheres. In F. O. Schmit & F. G. Worden (Eds.), The Neuroscience Third Study Program (pp. 5–19). MIT Press. Berggruen Essay Prize Competition, 2025, Consciousness Word Count: 6851

Sperry, R. W., Gazzaniga, M. S., & Bogen, J. E. (1969). Interhemispheric Relationships: The Neocortical Commissures; Syndromes of Hemisphere Deconnection. In P. J. Vinken & G. W. Bruyn (Eds.), Handbook of Clinical Neurology (pp. 273–290). North Holland Publishing Company.

Tye, M. (2003). Consciousness and Persons: Unity and Identity. MIT Press.

Weiskrantz, L. (2009). Blindsight: A Case Study Spanning 35 Years and New Developments. Oxford University Press.