Mary and her Sister: debunking a classic thought experiment

Some years ago I thought up an alternative version of Frank Jackson’s well-known thought experiment concerning Mary, the super-scientist who knows all the facts about the human visual system, the physics of light and colour, and so on, but has been brought-up in a black-and-white room and is therefore supposed to learn some new fact when she first sees a coloured object. I still have a soft spot for my version of Jackson’s story, at least as an intuition pump, and thought it worth bringing to the light of day.[1]


Mary has a twin-sister, Marianne, and both of them have been brought up in a black-and-white apartment by a group of scientific researchers. Unlike her sister, Marianne has not been educated in the sciences but has developed into an extremely talented musician. Neither sister has ever seen a colour – just black, white, and various shades of gray.

One day a member of the research team asks Marianne to take a break from her piano practice and presents her with a sheet of coloured paper. After Marianne has overcome her astonishment, the researcher asks her to describe her experience:

“Well, I was quite surprised at first, and a bit puzzled, but I guess that’s the effect of seeing something unexpected and new. I suppose I must be experiencing one of those ‘colours’ that my sister is always talking about. As for the colour itself, it’s hard for me to describe. I mean, I haven’t seen it before, and it’s unlike anything that I’ve ever experienced. It’s  … um … it’s just different from black, white or gray – that’s the only way I can describe it. Oh, hold on a minute, it seems to be quite dark, like a dark shade of gray … what I mean is, it’s nothing like white at all.”

So the researcher writes down in her report: ‘hard to describe; a colour I haven’t seen before; unlike black, white or gray; but dark like a dark shade of gray, and nothing like white’.

Now the researchers have already told Mary, the super-scientist, all the details of the colour that they are going to show to Marianne. They have asked Mary to predict what it would be like for her sister when she sees precisely that colour, and Mary has placed her prediction inside a sealed envelope. So after the researcher has left Marianne’s room, the envelope is torn open and on the sheet of paper inside is written:

“Hard to describe; a colour I haven’t seen before; unlike black, white, or gray; but dark like a dark shade of gray, and nothing like white.”

So Mary knew exactly what it would be like for her sister to experience seeing that particular colour!


What, then, did I conclude from this little story about Mary and her sister? Note, firstly, that no mention is made of which colour is on the sheet of paper. This avoids the unwarranted assumption on the part of the reader that what Marianne learns on first seeing the colour is along the lines of ‘what it is like to see red’ – which the reader will assume is the same as what it is like for them to experience seeing a red object. I am therefore putting the reader into a similar position, epistemologically speaking, to Marianne.

Secondly, Mary’s extensive knowledge of neuroscience, optics, and colour theory plays no part whatsoever in arriving at her prediction. In the original thought experiment, we are provided with a description of Mary’s knowledge base, and asked whether she could know in advance what it will be like for her when she first sees, say, a red object. So it is natural for the reader to assume that Mary must somehow use her specialized knowledge in order to understand what it would be like for her on the first occasion that she sees something red. In that sense, Jackson’s argument is a classic conjuring trick: the audience’s attention is misdirected towards Mary’s specialised knowledge, and away from the phenomenology. But if we describe the thought experiment correctly, then the phenomenology gives the game away: it’s just ‘a colour I haven’t seen before’.

So the point of my reformulation of Jackson’s thought experiment was to show that the knowledge argument gains a large part of its credibility from an implicit assumption that there is some phenomenal property, presumably an intrinsic property, which is common to all experiences that people have on seeing, say, a red object under normal conditions – a property that is common to my experience, to your experience, and to Marianne’s experience when she first sees such an object.

But this is an unwarranted assumption for an argument that aims to establish the impossibility of materialism – it effectively assumes at the outset what it wants to prove. It seems to me to be perfectly coherent for there to be no common property of the experiences of different people when they observe a red object – no property that is shared by my experience, your experience, and Marianne’s experience. Instead, the phenomenal property of my experience when I see a red object may be something like ‘a colour which I, Mike, recognize as that which I term red’), and this differs from the property of your experience when you see the same object (‘a colour which I, the Reader, recognize as that which I term red’), and from the property of Marianne’s experience when she observes a red object for the first time in her black-and-white room (‘a colour which I, Marianne, do not recognize’); such phenomenal properties would reflect the subjective nature of the experiences of separate individuals. To express this as a slogan: phenomenal properties should follow the phenomenology.

The alternative is to maintain that phenomenal properties are in some sense objective: that if you, and I, and Marianne in her black-and-white room, all look at some red object, then our experiences will all have the same phenomenal property. This would be to treat having an experience in the same manner as perceiving an external object: for example, if you and I and Marianne were all looking at a mountain, it would be the same mountain we were looking at even if we all perceived it from completely different angles and distances. But the mountain is an external object, and experiences are not.


[1] Section 4(c) of my article ‘The Possibility of Materialism’ sets out a more comprehensive response to Jackson’s thought experiment and the ‘knowledge argument’ against materialism.

Inner lives, or inner states?

Further comment on recent papers by David Chalmers (early draft of The Meta-Problem of Consciousness) and Keith Frankish (Illusionism as a Theory of Consciousness), highlighting another area where it is easy to talk at cross purposes.

Frankish rejects phenomenal properties, but nevertheless maintains that we have conscious experiences – if we take ‘experience’ in a functional sense, as referring to those mental states that are the direct output of our sensory systems (2016, p.13). But how are we to understand the word ‘functional’ in this context?

Let’s start by considering how Chalmers uses the word. He takes it that functional descriptions are essentially third-person descriptions that must be expressed in objective terms, such as those of observable behaviour. Consider, for example, the following passage from The Conscious Mind:

Given the premise that some explanandum is forced on us by first-person experience [it follows] that what needs to be explained cannot be analyzed as the playing of some functional role, for the latter phenomenon is revealed to us by third-person observation … (1996, p. 110, emphasis added)

Turning to ‘The Meta-Problem of Consciousness’ (2018), it seems that Chalmers wants to be especially clear on this point, because in that paper every single occurrence of the word ‘functions’ is immediately preceded by ‘behavioral’ (the one exception being a backward looking reference to ‘those functions’). And in discussing such functions, he refers to ‘reports’, ‘behaviors’, ‘consequences’, and such like.

Chalmers does acknowledge that there are internal mental states that are responsible for the behavior which we observe as part of a functional analysis, but these are not conscious states – they are psychological states that are defined in terms of their role in producing that behavior. They must, therefore, be clearly distinguished from any associated conscious experiences, even if the two are closely correlated.

What this means is that Chalmers limits people’s inner lives to their phenomenal experiences. A zombie, for example, has internal mental states that produce observable behaviour that is identical to that of a conscious human being – but it has no inner life, no consciousness whatsoever. Not even our own thoughts form part of our inner lives so far as Chalmers is concerned, since thoughts are mental states that can be defined in a functional manner. What does form part of our inner lives is our experience of having those thoughts.

This account would imply that when Frankish restricts ‘experience’ to some functional notion, he thereby entirely eliminates the first-person perspective. However, this does not appear to be what he has in mind, as can be seen from the following:

… zombies are presented as creatures very different from ourselves — ones with no inner life, whose experience is completely blindsighted. As Chalmers puts it, ‘There is nothing it is like to be a zombie… all is dark inside’ (Chalmers, 1996, pp. 95–6). And illusionists will not agree that this is a good description of us. Rather, they will deny the equivalence between having an inner life and having phenomenal consciousness. (2016, pp. 22-23, emphasis added)

These comments about ‘having an inner life’ are difficult to square with a desire to eliminate the first-person perspective. And Frankish goes on to say:

Introspection delivers a view of ourselves that is peculiarly vivid and compelling and that seems radically at odds with that of the physical sciences. It might give us access to an aspect of reality inaccessible to third-person science. … But it might merely give us an unusual perspective on the same reality … (2016, p. 24, second emphasis added)

On this picture, entirely different from that of Chalmers, there are mental states (very possibly complex representational states) that constitute experience. Such states can be understood scientifically thorough the behavior that they engender, their neurophysiological bases, and so on. But each of us also has an alternative, interior, perspective on those states, precisely because they are our own experiences.

This is a picture with which I largely concur. But a problem looms. No matter how we characterize mental states from the inner perspective – in qualitative terms or otherwise – it is difficult to see how there can be any a priori entailment that proceeds from our third-person scientific descriptions of such states, in terms of observable behaviour or neurophysiology, to our first-person descriptions of those same states from the inner perspective.[1] And that lack of a priori entailment  means that we are once again faced with an epistemic gap …

However, there is a straightforward reason for this lack of a priori entailment. The third-person viewpoint derives – as its name suggests – from the possibility of inter-subject agreement on facts, where that agreement can be obtained, at least in principle, from any person placed in the relevant situation. Hence, the third-person viewpoint is not to be thought of as a perspective on some objective realm, separate from another realm which is accessible only from a privileged first-person viewpoint; rather it is akin to a generalization of the first-person viewpoint. Now in introspection, my own mental states – as aspects of my inner life – are only cognitively available to myself. Hence, the criteria that I use for deciding whether a particular concept applies to my own experiences cannot conceivably be objective, third-person criteria. And this implies that there cannot be any a priori entailment from third-person descriptions to first-person descriptions.

But I do not believe that the materialist has any reason to fear this type of epistemic gap, which arises from the development of interiority within certain organisms. Even a zombie – if we accept that it has thoughts – will find an epistemic gap between its first-person descriptions and its third-person descriptions. However, an interesting question arises: might not the epistemic gaps noted by Chalmers – expressed in terms of explanation, conceivability, and knowledge (2010, pp. 104-111) – arise for the same reason? That is, could they be due to the distinction between first and third-person perspectives, rather than the qualitative aspects of consciousness?[2]

In ‘The Meta-Problem of Consciousness’ Chalmers takes Frankish’s rejection of phenomenal properties at face value. Rather curiously, however, he does not directly address the suggestion that each of us has an interior life nonetheless. But it is a moot question – can a strong illusionist really hold that we have inner lives, or have they in fact committed themselves to the view that we are zombies who merely have inner states?

References:

David Chalmers (1996) The Conscious Mind

David Chalmers (2010) The Character of Consciousness

David Chalmers (2018) ‘The Meta-Problem of Consciousness’ (version 2), draft paper available at https://philarchive.org/rec/CHATMO-32

Keith Frankish (2016) ‘Illusionism as a Theory of Consciousness’, Journal of Consciousness Studies 23:11–12, pp. 11–39

Notes:

[1] Could we rely here on the idea that both descriptions are of the same mental state? I do not think so. No matter how well motivated it is, this assumed identity is simply not relevant to the question as to whether there is any a priori entailment between the third-person and first-person descriptions – if you need the identity in order to show there is an entailment, that that entailment cannot be a priori. How concerned we should be about this lack of entailment is another matter, as I go on to discuss.

[2] This is a suggestion which I consider in more detail in section 4 of ‘The Possibility of Materialism’.

Phenomenal properties: do they even exist?

David Chalmers has made available an early draft of a new paper, ‘The Meta-Problem of Consciousness’, in which he addresses two topics. First, he sets out the meta-problem: why is it that we find explanations of phenomenal experience problematic? Second, he considers the claim that phenomenal consciousness is some form of introspective illusion, as exemplified in Keith Frankish’s recent paper ‘Illusionism as a Theory of Consciousness’ (see this interesting blog review by Peter Hankins).

Reading these papers by Chalmers and Frankish, one after the other, highlighted two areas where it seems easy to talk at cross purposes. I’ll start with phenomenal properties


The description of phenomenal properties that Chalmers provides in his earlier writings is quite straightforward: he refers simply to ‘those properties that characterize conscious states according to what it is like to have them’ and emphasizes that this definition involves no further substantive requirements, such as intrinsicality (2010, p. 104; see also 1996, p. 359). We might term this a minimal conception of a phenomenal property, and it is reflected in one of Chalmers’ comments in his recent draft paper: ‘To generate the hard problem of consciousness, all we need is the basic fact that there is something it is like to be us. We do not need further claims about intrinsicness, nonphysicality, and so on’ (2018, p. 31).

Of course, it is obvious from his writings taken as a whole that Chalmers believes that phenomenal properties are intrinsic and nonphysical. So why adopt this minimal definition? The reason is this: in order for his arguments against materialism to be effective, they must not assume at the outset some form of ontological divide between the phenomenal and the physical. As he says near the beginning of The Conscious Mind, ‘At this early stage, I do not wish to beg any questions about whether the phenomenal and the psychological will turn out to be the same thing’ (1996, p. 12). Instead, Chalmers takes conscious experience as a phenomenon that is in need of explanation: ‘The main intuition at work is that there is something to be explained—some phenomenon associated with first-person experience …’ (1996, p. 110).

It is precisely because he adopts this minimal definition of a phenomenal property that Chalmers is able to force the materialist between Scylla and Charybdis: either they account for those familiar epistemic gaps (of explanation, knowledge, conceivability), or else they have to reject conscious experience even as a phenomenon in need of explanation – which, suggests Chalmers, is to ‘deny the manifest’ (2010, p. 112).

For my part, I have no difficulty accepting phenomenal properties on this minimal conception, on the grounds that having an experience with (e.g.) a greenish aspect is different from having an experience with a reddish aspect, and that whatever it is that differs in respect of those experiences qualifies as a phenomenal property on Chalmers’ definition. I am therefore perfectly happy to refer to experiences as being characterized in qualitative terms – such as ‘greenish’ or ‘reddish’. However, I maintain that Chalmers’ definition of phenomenal properties does not ipso facto mean that experiences cannot be biological processes: one can surely characterize an experience as ‘reddish’ even if that experience turns out to be constituted by certain neurological activities in the brain.

Frankish, on the other hand, is having none of it … and in his article on illusionism he categorically repudiates the existence of phenomenal properties. Here’s one reason he gives (2016a, p. 26) for this rejection: ‘Phenomenal properties must not merely cause representations of phenomenality but have some genuinely “feely” aspect to them. And it is unclear what this could be. What phenomenal residue is left, once features such as privacy, intrinsicality, and ineffability have been stripped away?’ I can see the point here, but it doesn’t seem to me to be decisive. In particular, the illusion that experiences involve intrinsic, ineffable properties does still have what Frankish refers to as a ‘feely aspect’; for example, my experience when I look at the sky on a clear day surely has an aspect to it that I characterize as ‘bluish’. Renouncing even this does appear to open one up to the objection that one is denying the manifest.

I believe that for Frankish, the decisive reason for rejecting phenomenal properties in toto is that if we do not, then we will still face those familiar epistemic gaps, and the hard problem will bite us just as forcefully as it ever did:

But while I do not deny that we may be able to provide reductive accounts of many aspects of conscious experience, I doubt that these will be sufficient to justify realism about phenomenal properties in anything like the traditional sense. My worries centre on explanatory gaps. While identities may be initially inferred on the basis of partial explanations, we expect to be able to render them intelligible, giving reductive explanations of higher-level properties in terms of more basic ones. Why should consciousness be an exception, especially when the feature that resists explanation is such a central one? (2016b, p. 282)

While I am sympathetic towards Frankish’s uncompromising materialist attitude, I fear that his approach to phenomenal properties is problematic in two respects. First, as I have already noted, it leaves him wide open to the claim that he is denying the manifest. As Chalmers bluntly puts in his draft paper: ‘I think illusionism is obviously false, because it is obvious that people feel pain’ (2018, p. 35). Second, I believe that by attempting to neutralize the epistemic gaps, it diverts attention away from the real causes of those gaps (see The Possibility of Materialism’).

Of course, Frankish understands perfectly well that he needs to meet the objection that he is denying that consciousness exists. He attempts to do this by distinguishing between (i) phenomenal properties, and (ii) experience – a term which he says he will use ‘in a functional sense, for the mental states that are the direct output of sensory systems’ (2016a, p. 13, emphasis added). But this reference to ‘functional’ brings us to the second area where confusion needs to be avoided, which will require a separate post …

 

References:

David Chalmers (1996) The Conscious Mind

David Chalmers (2010) The Character of Consciousness

David Chalmers (2018) ‘The Meta-Problem of Consciousness’ (version 2), draft paper available at https://philarchive.org/rec/CHATMO-32

Keith Frankish (2016a) ‘Illusionism as a Theory of Consciousness’, Journal of Consciousness Studies 23:11–12, pp. 11–39

Keith Frankish (2016b) ‘Not Disillusioned: Reply to Commentators’, Journal of Consciousness Studies 23:11–12, pp. 256-289