In reading Jack Smart's excellent Stanford Encyclopedia article on the Identity Theory, I was again struck by the role that the distinction between type and token plays in philosophy of mind. This distinction was originally made by Charles Sanders Peirce back in (if memory serves) the 1870s (he also distinguished "tone", but that hasn't traveled well). I think it may even apply to biological taxonomy and the species problem.
Smart says that a type of mental state may, on some accounts, have different tokens. I won't get into this here (I'm a full blown identity theorist, and if folk psychological types fail to be identical to neurophysiological types, so much the worse for the categories of folk psychology). What I am interested in here is a more abstract issue. That is, can types and tokens always be distinguished?
Smart makes the following comment in passing:
If we continue to concern ourselves with first order properties, we could say that the type-token distinction is not an all or nothing affair. We could say that human experiences are brain processes of one lot of sorts and Alpha Centaurian experiences are brain processes of another lot of sorts. We could indeed propose much finer classifications without going to the limit of mere token identities.
So on his view, a type can be a token (the token of brain processes X that are terrestrial and not Alpha Centaurian). In short, types can have tokens that are types that can have tokens. This is amazingly familiar - it is Aristotle's logic of the diairesis, or classification by division. Aristotle says that if you take a type (a genos, or in Latin, a genus) that can be differentiated into inclusive members (eide, or in Latin, species) these can either be individuals (what Peirce calls "tokens") or they are themselves type terms in which case they are subaltern genera, or included general terms.
Types and tokens were, on Peirce's account, ideas or words. But now they have changed, as Aristotle's general and special predicates did, to become things. We have a strong concern in modern philosophy, and in particular in philosophy of mind and science, on "natural kinds". These are just types that exist independently of any ideas or words used to represent or name them. And this is the core issue when we talk about classification.
In the Linnaean taxonomy, the only genus was at one rank, and the only species at the rank below that. All species were part of a genus, and all genera contained at least one species. The species was the token of the generic type. Of course, Linnaeus arbitrarily (and self-consciously) set up several other ranks (five originally, and two others, phylum/division and family, were added later bringing the ranks to seven) in which each higher rank included the lower, so this is an attenuated form of Aristotle's diairesis. But only the species was regarded as a type term that only had tokens within it. In Aristotle's logic, latinized in the middle ages, this was called the infimae species, or the "lowest form". Those tokens were, of course, individual objects: living beings.
[Incidentally, the term "being" and cognates in other languages - être, ens, seiend - was commonly used in the traditional natural history to refer to organisms. When logicians, who were aware of naturalists' usages, discussed this, they often referred to "natural beings" or "organized beings" to distinguish the naturalists' use from the logicians'. Likewise, they would refer to "logical species" and "natural species" in order to make clear that they were not illicitly referring to one when they were talking about the other. Not all logicians - John Stuart Mill being the obvious example - made this clear, but Richard Whatley and others that followed him certainly did.]
But classification of the logical kind that takes natural kinds to be real things and not merely signs or concepts must allow for there being smaller kinds than the type in question. So for example every naturalist, and today every geneticist, knows that there are subaltern kinds within species. We talk these days about haplotype groups, subspecies, breeds, lines, and (you guessed) "types" within species all the time. Moreover, there can be types that have tokens in disparate types - the cytochrome C gene in a variety of more or less related species, or the amino acid guanine. Types can cross types.
So in the context of biology, Smart is exactly right. The type-token distinction is not an all-or-nothing affair. It is all over the place, so to speak. Classification depends very much on the property that is being tracked. And there are several ways to go about this.
Aristotle and his successors tended to classify groups of species in terms of their habitats - hence the whale and the crocodile were fishes, and the bat was a bird, because fishes lived in water and birds flew through the air. But Aristotle also classified in terms of the functional organs, blood type and so on. These days we also classify in terms of reconstituted relationships of descent, or phylogeny, between species and groups that are composed by them.
Peirce also wrote
In order that a Type may be used, it has to be embodied in a Token which shall be a sign of the Type, and thereby of the object the Type signifies. I propose to call such a Token of a Type an Instance of the Type.
which is something Aristotle would also have agreed with. Types (or Forms, as the philosophical literature translates eidos and species) must have instances to be "used" (Peirce, as the founder of pragmatism, thinks truth is the ability to be used, either practically or conceptually). So there is no genus without a species and no species without beings that are included under it.
And no form without a formed thing.
Anyway, this goes to the question of whether species are individuals (particulars) or not. A recent paper by Olivier Reippel in Cladistics revives this now-classic issue. If they are classes, then they can't be individuals, and vice versa. But I wonder if this is in fact the right contrast. Types, if they can be tokens, can also be individuals. It may not play well in some logical contexts of the modern day, or in, say, philosophy of language, but I do not see that it is so bad an idea. Maybe sets and classes are defined as the same thing (some would disagree anyway, as witness this post by Mark Chu-Carroll recently), but there is no reason why a class has to be something that is the same in all times and places, which is how the individualism thesis distinguishes classes from individuals (which are spatiotemporally restricted).
In Aristotle's logic, anyway, a "universal" was just a term or property that covered more than one individual. So a species may be a type-term or name, and still cover more than a single individual under that account. Perhaps there are reasons for not doing that in some kinds of logic, but it need not hold true all the time, as witness the discussion in Smart's article.
So a species may be a type with a name, and still be an individual in the spatiotemporally restricted sense. It may be instanced in its organisms, but have a geo-historical edge and duration. And there may be nothing to the type over and above the properties instantiated in those organisms. The type may not even have any shared properties, if we adopt, as Douglas Gasking argued back in 1960, a cluster view of biological types, as we almost certainly must. Recently, this view has been called "population thinking" and it, too, has been contrasted to the [supposedly] older class view of types, but really there is little reason to contrast them if "class" does not mean only "intensionally-defined set".
I don't know why I wanted to write all that. Probably just to get it out of my head for future reference.
Gasking, Douglas (1960), "Clusters", Australasian Review of Psychology 38:13-18.
- Log in to post comments
Types, if they can be tokens, can also be individuals. It may not play well in some logical contexts of the modern day, or in, say, philosophy of language, but I do not see that it is so bad an idea.
Neither do I. In computer science, pure "prototyping" programming languages do not have classes or types to serve as a template for an object. An object is it's own "type", and the concept of "type" takes on a much more subjective meaning.
There is a big difference between "type" in a common statically-typed programming language, and a biological species. A biological individual is not a member of species in the same way that a programming object is derived from a class. That's why I like prototyping languages - they're closer to biology.
Yes, hooray for prototype-based programming languages. Check out Io, if you haven't already (http://iolanguage.com/about/). It would be the best language in the world if it was a bit more mature (specifically, if it didn't keep changing).
IO is a cool language, but I should probably also take the opportunity to shamelessly plug my own prototyping language - Trans (www.transmuter.org). It was designed as a platform for experimenting with software evolution (especially ontogeny), and was inspired by biology and chemistry. Although I'm taking a temporary break from it right now, I must say that, when I was in my development fugue state, "type" never left the top 3 issues on my mind. I thought about it constantly, and I suspect many other language designers do as well.
For me, Type is defined by an objects properties, and in an evolutionary system, properties can evolve dynamically. Type is what an object "is" - it's ontological, and is not a fixed property of the object, as it is in common static programming languages. In the end, I came to the conclusion that type was ultimately subjective, and that it shouldn't be built in to the language, just as it isn't built in to evolution. In the continuum of our ancestry, I doubt there was one specific point or individual where we went from being a "reptile type" to a "mammal type". Prototyping languages can model evolution much more naturally and elegantly than fixed static languages do. Objects are simply derived from other objects with modification as needed - they are what they are, and do not belong to any fixed class. And if your classes are fixed, how can you evolve new ones?
This is, in fact, the late objection to Darwinian evolution - first by Agassiz (who denied that real species could transmute, so Darwin had to be saying there were no real species) and later by Catholic objectors to Haeckel, who said that a species was a logical type and therefore could not change.
You might investigate the logic of cladistics if you want to see a type-less classification - only specimens acting as prototypes in the construction of the tree are used. A specimen stands for any clade of which it is the sole member in that classification, from haplotype group up to very abstract levels of clades, depending on the interests and topics of the researcher.
Mr Wilkins wrote:
A philosopher whose knowledge of logic includes Peirce and Mill is well educated, if it includes Richard Whately then he/she is truely knowledgeable!
and later by Catholic objectors to Haeckel, who said that a species was a logical type and therefore could not change
In computer languages, this may correspond to compile-time types (logical types), and more dynamic run-time types which are harder to define and depend on the context, or as you put it "how instances are used".
Damn! You know it never occurred to me to see evolution as a type-redefinition process...
Type redefinition or type-transmutation? Although.. redefining your own species at will might be a useful adaptation :)
Peirce's notion of tone, while a poorly chosen word (and he tried others) is a pretty valuable concept. After all we want to be able to distinguish between a very particular instance as a sign - that word written there - and then variants of this - say our abstract type of a written word. Clearly the meaning of the word is signified by a particle instance of the word written in say a newspaper. But also clearly we can think of the word in more abstract ways that also signifies the meaning of the word.
It takes a while to wrap your mind around it but once you do it's really quite valuable.
One should also note that Peirce was a scholastic realist. That is he thought abstract ideas were independent of any particular mind thinking about them. This is quite different from many folks working in philosophy of mind who tend to be more nominalistic - although arguably the notion of a meme is a bit of a throwback to Peirce. Anyway because of Peirce's realism his token - tone - type trichotomy ends up being a bit more valuable than how it's transformed itself in some uses.
To add - for the comp sci relevancy once you have classes that have shared variables with values then you really are in a middle ground between a true abstract class and an instance. So you could argue that Peirce's tone is an anticipation for this need for a middle ground.
Speaking of theory of mind, it would be great that in a future post you epressed your opinion on the ideas in, or reviewed Douglas Hofstadter's latest book. I haven't read it, I just read the American Scientist review of it. Some of the ideas are quite disturbing. I quote from the review:
There are several chapter's on Goedel's proof. Hofstadter argues that this proof shows how meaningful self-reference ("strange loops") can emerge out of so apparently unpromising base as the dry -and semantically empty- logic of Bertrand Russell and Alfred North Whitehead. Why should we care about that? Because, he says, the particular way in which Goedel's proof went beyond Russellian logic is essentially the same as the way in which psychology goes beyond neuroscience, or mind beyond brain
[...]
Well, remember the antireductionist claim cited above: that the self is -repeat, is- an abstract pattern, which emerges from a feedback system of sufficient complexity -namely, the adult human brain. If the self, the mind or the soul -Hofstadter uses these three terms more or less interchangeably- is not the brain, it's not obvious that it must cease to exist when the (dead) person's brainstuff is dispersed by flames or decay.
What I found most disturbing is that the reviewer, Margaret A. Boden, doesn't seem to be disturbed by these ideas...
Hofstadter used to be an intelectual hero for me (I love Bach!), but this is quite disappointing.
Clark, thanks for the comment. I love Peirce not in a small way because he is a scholastic (and not merely a scholastic realist, as you correctly identify him). Pretty well all his terms and ideas are derived from the late scholastics, and in particular the "Nominalists". IMO...
Ribozyme, I try to be incompetent in as few fields as possible, and while I am happy to be incompetent in the philosophy of biology, I am rather more circumspect about being incompetent in philosophy of mind or cognitive science (which, for my sins, I teach this coming semester). But Margaret Boden is one of my intellectual heroes, or was 15 years ago when I read her, so if she says a view is acceptable, I'm going to try hard to find a charitable interpretation in which she is right.
Patterns however, do not exist unless there is a substrate for them to exist in, and dynamic patterns do not exist unless the substrate is active in the right ways. If mind is an abstract pattern, then it had better be a pattern in a thinking machine to continue after death...
The substrate is also a pattern, so don't we have a problem of infinite recursion? And can a "substrate" exist if there is not at least one "pattern" in it? Types and tokens...
Ach! Now you're asking me to do real philosophy, you swine!
Substrates have patterns, but they are not themselves patterns. My personal view is that there are as many patterns as there are things to differentiate in a substrate. Or, in less falutin' words, pattern is relative to what interests you to observe...
Well, let me put it this way - if "mind" has neurons as a substrate, and neurons have biochemistry as a substrate, and biochemistry has atomic physics as a substrate, and atomic physics has subatomic physics as a substrate, and then quantum mechanics, and then...? (reductionism). Is it turtles all the way down forever and ever? And of course in the other direction, patterns may also have mind as a substrate ("memes"), and other patterns may depend on memes (i.e. our technology), and... more turtles.
Or..perhaps mind is a crucial link in the system (though not the entire system). Can the substrate exist in the first place if there's no mind to perceive it?
Well I am inclined to say that the substrate exists independently of what we conceive it to be, that being rather in the nature and definition of "substrate". There has to be a "pattern bearer", so to speak. But at what level you cease having subaltern patterns or substrates, well, I don't know. For me, something subatomic is low enough.
As for mind having patterns as a superstructure, I deny this. Mind just is that having of certain types of dynamic patterns. Memes don't exist except as aspects of mind, and mind is at least constituted by memes (that is, memes don't exist on their own, and they comprise some of mind).
I understand this - biochemistry, subatomic physics, etc are "types" that we subjectively confer upon reality, and reality is under no obligation to recognize them or our science. It is was it is. But is this distinction you're making between pattern and substrate not also a subjective definition of types? Wouldn't one type be less confusing - mind, or bruce, or whatever?
I didn't say that mind has a superstructure - only that memes may have mind as a substrate (if you accept the existence of the same meme in more than one mind).
John: I thought that what I quoted from the review of Hofstadter's new book would be enough to yield the gist of the review, but I think some more needs to be added:
Hofstadter does propose a substrate, but what he proposes is, in my opinion (and I'm not a philosopher), quite ludicrous. Part of the book (according to the review) is about the rather recent death of Hofstadter's wife and the grief it brought to him. It seems to me that he is clutching at straws in a rather misguided attempt to come to terms with that. I think it is the assumption that the mind is more than brain functions ("psychology is more than neuroscience") that leads Hofstadter to this ridiculous conclusions.
Most of the comments in this post are quite obviusly from philosophers, or at least from people who are well versed on modern philosophy. I'm just a scientist, quite ignorant on the subject. I'm now very curious about Margaret Boden, after you describe her as a hero of yours. In your words, what are her contributions in her field (cognitive science, from what I read) that make her so?
I should also add that the "mind" type can be divided into many different sub-types, such as functional intelligence, functional consciousness, and phenomenal consciousness (qualia). The last is the "hard problem of consciousness" and the most difficult to explain in terms of a substrate due to its subjectivity, although some philosophers (Dennett?) don't believe it actually exists.
If you google "artificial consciousness", you'll find the suggestion that a thermostat may be "conscious", considerably expanding the "mind" type. To me, that's almost panpsychism (the inverse of solipsism?)