Archive for June, 2014

nostrilsSmell is the most elusive of the senses. Sight is beautifully structured and amenable to analysis in terms of consistent geometry and a coherent domain of colours. Smells… how does one smell relate to another? There just seems to be an infinite number of smells, all one of a kind. We can be completely surprised by an unprecedented smell which is like nothing we ever experienced before, in a way we can’t possibly be surprised by a new colour (with some minor possible exceptions). Our olfactory system effortlessly assigns new unique smell experiences to substances that never existed until human beings synthesised them.

There don’t even seem to be any words for smells: or at least, the only way we can talk about them is by referring to “the smell of X”, as in a “smoky smell” or “the smell of lemons”. We don’t have to do that to describe shapes or colours: they can be described as “blue”, or “square” without our having to say they are “sky-coloured” or “the shape of a box”. (Except perhaps in the case of orange? Is “orange” short for ‘the colour of oranges’?) Even for taste we have words like “bitter” and “sweet”. The only one I can think of for smells is “nidorous’, which is pretty obscure – and in order to explain it I have to fall back on saying it describes the “smell of” burning/cooking meat. All we have to describe smells is “strong” and “faint” (my daughter, reading over my shoulder, says what about “pungent”? She does not consider “pungent” to be merely a synonym of “strong” – you may be indifferent to a strong smell, but not to a pungent one, she claims).

With that by way of preamble, let me introduce the interesting question considered here by William Lycan: does smell represent? When we smell, do we smell something? There is a range of possible answers. We might say that when I smell, I smell sausages (for example). Or that I smell a smell (which happens to be the smell of sausages). Or I might say I just have a smell experience: I may know that it’s associated with sausage smells and hence with sausages, but in itself it’s just an experience.

Lycan (who believes that we smell a gaseous miasma) notes two arguments for something like the last position – that smell doesn’t represent anything. First, introspection tells us nothing about what a smell represents. If I were a member of a culture that did not make sausages or eat meat, and had never experienced them, my first nose-full of sausage odour would convey nothing to me beyond itself. It’s different for sight: we inherently see things, and when we see our first sausage there can be no doubt we are seeing a thing, even if we do not yet know much about its nature: it would be absurd to maintain we were merely having a visual experience.

The second argument is that smells can’t really be wrong: there are no smell illusions. If a car is sprayed with “new car” perfume to make us think that it is fresh off the production line, we may make a mistake about that inference, but our nose was not wrong about the smell, which was real. But representations can always be wrong, so if we can’t be wrong, there is no representation.

Lycan is unimpressed by introspective evidence: the mere fact that philosophers disagree about what it tells us is enough, he feels, to discredit it. The second argument fails because it assumes that if smells represent, they must represent their causes: but they might just represent something in the air. On getting a whiff of my first sausage I would not know what it was, but I might well be moved to say “What’s that appetising (or disgusting) smell?”  I wouldn’t simply say “Golly, I am undergoing a novel olfactory experience for some opaque reason.”  I think in fact we could go further there and argue that I might well say “What’s that I can smell?” – but that doesn’t suit Lycan’s preferred position. (My daughter intervenes to say “What about ‘acrid’?”)

Lycan summarises a range of arguments (One is an argument by Richardson that smell is phenomenologically “exteroceptive”, inherently about things out there: Lycan endorses this view, but surely relying on phenomenology is smuggling back in the introspection he was so scathing about when the other side invoked it?). His own main argument rests on the view that how something smells is something over and above all the other facts about it. The premise here is very like that in the famous thought experiment of Mary the colour scientist, though Lycan is not drawing the same conclusions at all. He claims instead that:

I can know the complex of osphresiological fact without knowing how the rose smells because knowing is knowing-under-a-representation… that solution entails that olfactory experience involves representation.

That does make some sense, I feel (What about “osphresiological”! we’re really working on the vocabulary today, aren’t we?). You may be asking yourself, however, whether this is a question that needs a single answer. Couldn’t we say, yes sometimes smells represent miasmas, but they can also represent sausages; or indeed they can represent nothing.

Lycan, in what I take to be a development of his view, is receptive to the idea of layering: that in fact smells can represent not just a cloud of stuff in the air, but also the thing from which they emanated. That being so I am not completely clear why we should give primacy to the miasma. Two contrary cases suggest themselves. First, suppose there is a odour so faint I don’t even perceive it as such consciously, but have a misty sense of salsiccian (alright, I made it up) presence which makes me begin to think about how agreeable a nice Cumberland sausage for lunch might be. Wouldn’t we say that in some sense the smell represented sausages to me: but we can’t say it represented a miasma because no such thing ever entered my mind?

Second, if we accept layering we might say that the key point is about the essential or the minimal case: we can smell without that smell representing a sausage, but what’s the least it can represent and still be a smell? Can it represent nothing? Suppose I dream and have an odd, unrecognisable experience. Later on, when awake, I encounter a Thai curd sausage for the first time and find that the experience I had was in fact an olfactory one, the smell of this particular kind of comestible. My dream experience cannot possibly have represented a sausage, a miasma, a smell, or anything but itself because I didn’t know what it was: but, it turns out, it was the smell of curd sausage.

I think your reaction to that is likely to depend on whether you think an experience could be a smell experience without being recognisable as such; if not, you may be inclined to agree with Lycan, who would probably reiterate his view that smells are sensing-under-a-representation. That view entails that there is an ineffability about smell, and Lycan suggests this might help account for the poverty of smell vocabulary that I noted above. Interestingly it turns out that this very point has been attacked by Majid and Burenhult, albeit not in a way that Lycan considers fatal to his case. Majid and Burenhult studied the Jahai, a nomadic hunter-gatherer tribe on the Malaysian peninsula, and found that they have a very rich lexicon of odour terms, such as a word for “the smell of petrol, smoke and bat droppings” (what, all of them?). It’s just us English speakers, it seems, who are stuck with acrid nidors.

Turing surprisedSo, was the brouhaha over the Turing Test  justified? It was widely reported last week that the test had been passed for the first time by a chatbot named ‘Eugene Goostman’.  I think the name itself is a little joke: it sort of means ‘well-made ghost man’.

This particular version of the test was not the regular Loebner which we have discussed before (Hugh Loebner must be grinding his teeth in frustration at the apprent ease with which Warwick garnered international media attention), but a session at the Royal Society apparently organised by Kevin Warwick.  The bar was set unusually low in this case: to succeed the chatbot only had to convince 30% of the judges that it was human. This was based on the key sentence in the paper by Turing which started the whole thing:

I believe that in about fifty years’ time it will be possible, to programme computers, with a storage capacity of about 109, to make them play the imitation game so well that an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning.

Fair enough, perhaps, but less impressive than the 50% demanded by other versions; and if 30% is the benchmark, this wasn’t actually the first ever pass, because other chatbots like Cleverbot have scored higher in the past. Goostman doesn’t seem to be all that special: in an “interview” on BBC radio, with only the softest of questioning, it used one response twice over, word for word.

The softness of the questioning does seem to be a crucial variable in Turing tests. If the judges stick to standard small talk and allow the chatbot to take the initiative, quite a reasonable dialogue may result: if the judges are tough it is easy to force the success rate down to zero by various tricks and traps.  Iph u spel orl ur werds rong, fr egsampl, uh bott kanot kope, but a human generally manages fine.

Perhaps that wouldn’t have worked for Goostman, though because he was presented as relatively ignorant young boy whose first language was not English, giving him some excuse for not understanding things. This stratagem attracted some criticism, but really it is of a piece with chatbot strategy in general; faking and gaming is what it’s all about. N0-one remotely supposes that Goostman, or Cleverbot, or any of the others, has actually attained consciousness, or is doing anything that could properly be called thinking. Many years ago I believe there were serious efforts to write programs that to some degree imitated the probable mental processes of a human being: they identified a topic, accessed a database of information about it, retained a set of ‘attitudes’ towards things and tried to construct utterances that made sense in relation to them.  It is a weakness of the Turing test that it does not reward that kind of effort; a robot with poor grammar and general knowledge might be readily detectable even though it gave signs of some nascent understanding, while a bot which generates smooth responses without any attempt at understanding has a much better chance of passing.

So perhaps the curtain should be drawn down on the test; not because it has been passed, but because it’s no use.

gerald edelmanGerald Edelman has died, at the age of 84. He won his Nobel prize for work on the immune system, but we’ll remember him as the author of the Theory of Neuronal Group Selection (TNGS) or ‘Neural Darwinism’.

Edelman was prominent among those who emphasise the limits of computation: he denied that the brain was a computer and did not believe computers could ever become conscious…

In considering the brain as a Turing machine, we must confront the unsettling observations that, for a  brain, the proposed table of states and state transitions is unknown, the symbols on the input tape are ambiguous and have no preassigned meanings, and the transition rules, whatever they may be, are not consistently applied. Moreover inputs and outputs are not specified by a teacher or a programmer in real-world animals. It would appear that little or noting of value can be gained from the application of this failed analogy between the computer and the brain.

He was not averse to machines in general, however, and was happy to use robots for parts of his own research. He drew a distinction between perception, first-order consciousness, and higher-order consciousness; the first could be attained by machines we could build now; the second might very well be possible for machines of the right kind eventually – but there was much to be done before we could think of trying it. Even higher-order consciousness might be attainable by an artefactual machine in principle, but the prospect was so remote it was pointless to spend any time thinking about it.

There may seem to be a slight tension here: Turing machines are ruled out, but machines of another kind are ruled in. Yet the whole point of a Universal Turing Machine is that it can do anything that any machine can do?

For Edelman the point was that the brain required biological thinking, not just concepts from physics or engineering. In particular he advocated selective mechanisms like those in Darwinian evolution. Instead of running an algorithm, the brain offered up a vast range of neuronal arrays, some of which were reinforced and so survived to exert more influence subsequently. The analogy with Darwinian evolution is not precise, and Francis Crick famously said the whole thing could better be called ‘Neural Edelmanism’ (no-one so bitchy as a couple of Nobel prize-winners).

Edelman was in fact drawing on a different analogy, one with the immune system he understood so well. The human immune system has to react quickly to invading infections, synthesising antibodies to new molecules it has never encountered before; in fact it reacts just as effectively to artificial molecules synthesised in the lab, ones that never existed in nature. For a long time it was believed that the system somehow took an impression of the invaders’ chemistry and reproduced it; in fact what it does is develop a vast repertoire of variant molecules; when one of them happens to lock into an invader it then reproduces vigorously and produces more of itself to lock into other similar molecules.

This looks like a useful concept and I think Edelman was right to think it has a role to play in the brain: but working out quite how is another matter. Edelman himself built a novel idea of recategorisation based on the action of re-entrant loops; this part of the theory has not fared very well over the years. The NYT obituary quotes Gunther Stent who once said that as professor of molecular biology and chairman of the neurobiology section of the National Academy of Sciences, he should have understood Edelman’s theory – but didn’t.

At any rate, we can see that Edelman believed that when a conscious machine was built in the distant future it would be running a selective system of some kind; one that we could well call a machine in everyday terms, though not in the Turing sense. He just might be vindicated one day.