Science Friction
Rachel at A Historian’s Craft and Kevin at Civil War Memory have both been thinking about how much historians should think about philosophy. Although they take different positions on the issue, they both approach it in a refreshingly un-polemical fashion (contrast with the “that’s you that is” pettiness of this embarrassing exchange between Alun Munslow and Arthur Marwick). It’s almost inevitable that the p-word comes up, but it’s interesting that the word “postmodernism” seems to be used more often by people who are against it than people who are for it, whatever it is. Too often it seems to be a label attached to a conflation of lots of different (and not always compatible) theories, but let’s stick with the stereotypical view of postmodernism for now. Here are two recognisable stereotypes:
The traditional empiricist, who believes that what historians do is to scientifically examine archival evidence to find out what really happened in the past, something which is achievable if you eliminate bias.
The postmodernist who believes that everything is culturally constructed, that an objective scientific study of the past is impossible, and that even science itself is an ideologically suspect paradigm.
Whether these stereotypes are true or not (and you should always be suspicious of stereotyping – isn’t it funny how stereotypes are always someone else?) they crudely illustrate what I’m trying to get at in this post: that both extremes in the postmodernism wars seem to have a stereotypical and inaccurate view of science.
First of all I have to say that most scientists would laugh at the idea that history can be scientific, but that seems so obvious that it’s hardly worth discussing. The more interesting point is that science undermines some of the assumptions of traditional empirical history and supports some points of view which might be characterised as “postmodern”.
Chris at Mixing Memory says: “Spend a little time in a cognitive science lab, and you will quickly be disabused of any inclinations towards naive realism.” Spend some time reading Mixing Memory and you’ll find plenty of empirical scientific evidence that human perceptions can be very unreliable (even before conscious bias has any chance to act on them) and that language can influence perception. In this post Chris discusses linguistic relativity and looks at an experiment which appears to show that the colour terms in a language influence how people perceive colour. He points out that colour perception is one of the hardest areas in which to prove linguistic influence, but that the influence of language on concepts such as gender is much less controversial.
Language isn’t just problematic because of its insidious influence on perception and conceptualisation. The biggest problem is meaning. Post-structuralist theory (often conflated with postmodernism but not quite the same thing) suggests that meaning does not reside in text and cannot be fixed. This is a controversial idea to some people, but I don’t see many convincing arguments against it. So far science has been unable to even define what meaning is, let alone where it comes from. Edmund Blair Bolles grapples with the problem at Babel’s Dawn (here and here), a blog devoted to investigating the evolutionary origins of human speech. His mention of Searle’s “Chinese room” is particularly interesting here. Searle has been criticised for failing to define “understanding” but I don’t blame him for that because nobody really knows. Whatever it is, his thought experiment convincingly demonstrates that computers do not understand anything in the way that humans do. All the work that has gone into Artificial Intelligence has failed to come up with a machine which can understand the meaning of human language. Even passing the Turing test (which only requires creating the illusion of talking to a real person, not actual understanding) is surprisingly difficult. If language is as straightforward as the opponents of post-structuralism believe, then shouldn’t it be easy for machines to understand?
Nothing actually is what it’s called. Science can give us some incidental illustrations of this principle at work. Not too long ago, astronomers decided to change their definition of “planet” so that it no longer included Pluto. These astronomers clearly recognised that they are not just neutrally discovering what’s out there. Instead they are classifying things according to an arbitrary taxonomy which they have constructed themselves. If that taxonomy turns out not to be very useful for its intended purpose then it can be changed. But we all know that that’s not how the move was interpreted outside astronomy. Some people got quite emotional about it – Pluto is a planet because everyone knows it is!
When he laid the foundations of information theory, Claude Shannon was careful to separate information from meaning. A mathematician and engineer writing equations about the transmission of signals might be a prime suspect for being a realist/reductionist/whatever, but thanks to the separation of meaning from information Shannon’s theory is compatible with structuralism and post-structuralism.
But the real point I want to make about information theory is the way it undermines reconstruction of the past. (A lot of what I know about this come from Andrew Hickey, who explains it here) Complete information allows you to reconstruct the original message exactly as it was. In Shannon’s calculations the message was either a series of characters drawn from a finite set, or a continuous signal, such as a sound wave. If the message we want to reconstruct is the reality of the past then we clearly don’t have enough information to reconstruct anything. At best, historical sources (whether they’re documents, photographs, films, sound recordings) only capture some aspects of reality (and remember that reality is usually filtered through people’s perception and cognition, with all the problems that cognitive science has identified).
Another thing I learned from Andrew Hickey is Ashby’s first law of cybernetics, which he sums up as: in order to be in control of something, you need more available options than possible outcomes. This has major implications for historical causation. Think of how many possible outcomes there are when ruling a country or commanding an army. Can we really talk about anyone in history being in control of anything? This is one of the reasons why I’m dissatisfied with the explanations offered by most historians. Should we be talking about “influence” rather than “command” or “control”?
Chaos theory makes things even more difficult. In a chaotic system, small changes in the initial conditions can produce disproportionately large difference in the outcome. These systems can be difficult to model accurately because of their complexity, which is a major problem for counter-factual thinking. The consequences of something in history happening slightly differently might be wildly unpredictable.
This post is a fairly superficial introduction to things I don’t really understand very well (and I haven’t even gone into quantum theory, where things get really crazy) but if I’m even half right you’d have to ask why an extreme empiricist would cling to science and an extreme postmodernist would reject science. Maybe they don’t. After all, I did start with some unrealistic stereotypes. Nevertheless, it should now be clear that we do have some big problems here. Can we just ignore them and get on with doing history? If not, how much time should we spend thinking about them? I’ll be thinking about that in the next post.
- C E Shannon, ‘A mathematical theory of communication’, Bell System Technical Journal, 27 (1948), pp. 379-423, 623-656.
Comment by Brett — 6:31 am, 10 June 2007 [permanent link to this comment]
When I first started history (coming from science) I was all fired up to do battle with the evil postmodernists infesting the academy … I must confess I’ve never met anyone who has come close to matching the stereotype I had in my head. People are generally more pragmatic than that, they operate somewhere in the middle between the two extremes.
Yes. My reaction to this is to note that this is true of all of our interactions with reality, whether we are talking about history or not. But just as we don’t let that lead us into paralysis (“well, I can’t know for sure that there’s a truck about to run over me, so I may as well just stand here”), it shouldn’t lead to paralysis in historical inquiry (which I’m sure is not your conclusion!) We can’t reconstruct the past completely, that’s all, but we already knew that :)
Again, I agree — but didn’t we already know this? As Moltke the elder said: no battleplan survives contact with the enemy. Reality is messy. So whenever I read that X caused Y, I know it’s only a partial (if perhaps dominant) cause and that there are others which could be added. Do you think that’s not how it’s meant to be understood?
Probably best to steer clear of QM altogether. For one thing people are too big to exhibit quantum behaviour, so applying it to history is going to be extremely, er, problematic!
Comment by Gavin Robinson — 8:49 am, 10 June 2007 [permanent link to this comment]
Yes, I think the whole point of this post is that it’s supposed to point out some obvious things in an uncontroversial way that most people should be able to relate to without going into the dreaded French philosophy. That way anyone who does fit the extreme empiricist stereotype would have to admit that their position was self defeating. Either that or become so enraged that they explode. But at most only a very small minority are anything like those stereotypes. My experience has been the same as yours: most people just get on with it using whatever methodology they find most useful for their projects and don’t feel threatened by people who do things differently. I mentioned in one of my earliest posts that I spent most of the 1990s studying history in universities and remained completely unaware of the theory wars!
I’ll be continuing the discussion in a future post, but for now I’ll say that I think Kevin Levin is right that while historians might be affected by big philosophical/scientific problems, we clearly aren’t in a position to solve them ourselves. And obviously if I thought we should just give up on history then I wouldn’t be writing this!
Maybe I’ve misunderstood the rules of the game about causation, which at the very least suggests historians need to discuss their methodology in more detail so that people don’t get the wrong end of the stick. The impression that I get is still that some historians oversimplify causation by making one factor appear to be more important than all the others. But maybe that impression is partly caused by the way historians oversimplify each others’ arguments when they’re disagreeing with each other.
From what little I know about quantum theory, some of the hypotheses which have been thought up to explain the unexpected behaviour of waves and particles under certain conditions have major implications for historical causation, and for reality itself. If many worlds is true then anything that can possibly happen will happen in some universes, but not in others. If hidden variable is true then the past is determined by the future. But they might not be true.
Comment by Gavin Robinson — 4:57 pm, 10 June 2007 [permanent link to this comment]
And news just in: Chris at Mixing Memory posted about an experiment which provides empirical proof of the truth effect at work. Non-experts are more likely to accept bad explanations of psychological phenomena if they include irrelevant neuroscience terminology. I’d love to see some similar experiments on history.
Comment by Brett — 6:25 am, 11 June 2007 [permanent link to this comment]
Well, maybe I’ve misunderstood the rules of the game, etc! Which rather goes to your point that meaning does not inhere in text (oh dear, am I being post-structuralist? I may have to have a lie down to recover) I think it’s this sort of ambiguity that’s the cause (uh-oh) of so much of the misunderstanding between empiricists and postmodernists. Take the idea of social construction, for example. There’s nothing that will get the back of a physical scientist up quicker than to tell them that “science is socially constructed”. In return, a postmodernist would laugh at the retort that “science is an objective source of knowledge about the world”. But both of these are extreme statements that neither will fully subscribe to, if pressed. A moment’s reflection upon the way that science has developed historically (even the very fact that it HAS a history) and the way it is done in practice shows that it is indeed a social construction. But that doesn’t mean that it’s only a social construction, because it obviously correlates with and describes an external reality more or less accurately, such that we can manipulate that reality in non-trivial ways (see, eg, the Manhattan Project). So if instead, we all agreed that “science is partly socially constructed AND partly objective”, there’d still be a lot left to argue about, but we’ve already conceded that each rhetorical extreme has a point. Similarly with meaning and texts … that meaning isn’t fixed doesn’t mean that all readings of a text are equally valid. And so on.
Well, that was all a lot clearer in my head! Coincidentally, the comments on a recent post at Galactic Interactions (more so than the post itself) seems to cover a lot of this same ground.
Just on that last point, many worlds is more metaphysics than anything else, IMHO, while most physicists would accept that Bell’s inequality theorem does away with hidden variables (it doesn’t entirely, but then you have to make further assumptions and the end result isn’t necessarily any prettier). So I’d be reluctant to draw any firm conclusions from them. But even if you could, I’m not sure where it would get you. That all possible events happen somewhere in the parallel universes doesn’t really help us understand the history of this universe any better, it seems to me: we can talk about contingency and probability of events already. With the future determining the past — I assume you’re referring to advanced waves? — whether that’s true at a quantum level or not, people don’t act like it is, their past determines their future. Why this should be — why time has an arrow — is actually quite mysterious and very interesting indeed. Probably, more historians ought to be aware of the second law of thermodynamics, as it’s quite fundamental to their job!
Maybe what I’m really saying is that you should talk about QM after all, as it’s a lot of fun :)
Pingback by Investigations of a Dog » That would be an ecumenical matter — 1:48 pm, 12 June 2007 [permanent link to this comment]
[...] the discussions at A Historian’s Craft and Civil War Memory about history and philosophy. In that post I took some of the philosophical problems that affect history and tried to restate them in [...]
Comment by Gavin Robinson — 2:11 pm, 12 June 2007 [permanent link to this comment]
When I said that I’m not convinced by any arguments against post-strucuralism, I should also have said that I’m not convinced by any arguments in favour of the strong version of post-strucuralism which seems to imply that language is completely free floating. The logical conclusion of that would be that communication is impossible, but experience suggests that it isn’t. I suspect that the debate over whether language is fixed or free floating is a false dichotomy. I don’t think it’s fixed or free floating – it’s somewhere in between, or maybe something completely different. A language that never changes will become useless, but a language the changes too quickly and unpredictably will also become useless. We seem to be able to understand language without trying too hard, but we don’t know really know how we do it. Maybe that’s not a problem for most people – you don’t need to know how to build an engine to be able to drive a car – but it is a problem for academics in at least some disciplines.
What you said about science also goes for historical narratives. In this case the suggestion that historians neutrally write down the facts of how the past really was, and the suggestion that they construct narratives which are no more true than fiction are the two extremes, but they’re not mutually exclusive. Historical writing is a form of narrative which has genre conventions, historians do arbitrarily construct their own narratives about the past, the facts could be arranged and interpreted differently to tell a different story, but none of that automatically makes it fiction. Narrative structures, rhetorical strategies, and the “truth effect” aren’t inherently bad: they can be used for good or evil depending on the historian. In the wrong hands they might lead people to believe something that’s false, but in the right hands they can reinforce something that’s true.
When I say “true” and “false” i don’t mean to imply a binary opposition. I tend to think that truth is more of a continuum. Absolute truth is unattainable, but some things are more true than others. I think the word “fact” can be very misleading, but I can’t avoid using it because it’s so much more convenient than “epistemic probability”.
Pingback by Investigations of a Dog » Information vs Meaning: A False Dichotomy? — 2:33 pm, 2 July 2007 [permanent link to this comment]
[...] Claude Shannon, the father of information theory) and some of its implications. For example, in this post I pointed out that Shannon’s separation of meaning and information is compatible with [...]