Lo, Quine! 
In Theories and Things and Perspectives on Quine, Quine defines an observation sentence for an individual in this way:
If querying the sentence elicits assent from the given speaker on one occasion, it will elicit assent likewise on any occasion when the same total set of receptors is triggered; and similarly for dissent. [PoQ, p.3]
Yet this seems obviously inadequate. Consider and individual asked to watch a series of images and count the number of rabbits that appear. Obviously, "That is the first rabbit" ought to count as an observation. Nevertheless, the very same image and so the very same receptor trigger will not elicit assent on subsequent occasions. Rather, when the subject sees the same picture she will say "That is the second rabbit."

One might amend the definition so as to include context; for example, the agent will assent when the same receptors are triggered and the neural machinery responsible for memory is active in the same way. This seems inadequate for several reasons. First, the revised version individuates occasions so specifically that they might be unrepeatable. This would trivialize the consequent of the conditional. Second, the revised version makes remembering something count as an observation. Third, there may be no clear line between the neural machinery of memory and that of calculation. That would make arithmetic count as observation.

Similar problems arise for sentences like "It is cold." My sensitivity to cold depends in part on my state of mind and attention. Assent requires more than just a specific temperature on my receptors.

In addition to the specific objection, I disagree with Quine's general fixation on stimulations at the sensory periphery. Perception and memory are richly interwoven with the environment; cf. d-cog.

[ add comment ] ( 4430 views )   |  [ 0 trackbacks ]   |  permalink
Watch Thag simulate the world 
Scholars typically explain the demise of the Neanderthals by claiming that they were better suited for colder climates and so died out when the Ice Age ended. They had bigger brains than Homo sapiens, however, and so they probably could have still out-thought us. Their cleverness suggests other possibilities.

Cristyn is working on a sound installation inspired by the study of Neanderthals, and a couple of days ago we had a conversation that involved posing this scenario:

Imagine Homo sapiens had died out, and the Neanderthals had lived on to develop an advanced technological civilization. They would be curious as to what the world would have been like had they died out. So a sufficiently advanced Neanderthal civilization would run simulations in which we modern humans developed and lived our lives.

The lack of a satisfactory explanation for why Neanderthals died out is just what we would expect if our world were such a simulation. The Neanderthal high programmer would simply write in the removal of their Neanderthal forebears. Deus ex machina.

To jump to a conclusion: We have no direct way of knowing if our world is a simulation or not. As such, we have good reason to believe that either (a) we are living in a Neanderthal's simulation of a world, or (b) there is a complete naturalistic explanation of why Neanderthal's died out. The radical contingency of natural selection means that we would never have decisive reason to believe the latter rather than the former.

Of course, this is a variation of the Simulation Argument. However, it ups the stakes a bit: The simulation argument admits that our modern human civilization developed, but asks if we are in it or in a simulation of it. The new argument leaves us with the possibility that there never was really a civilization like ours. Perhaps Napolean, the Sears tower, the TV show Star Trek, and the moon landing were never anything more than an idle-time process on a powerful computer run by a caffeine-addled trans-neanderthal IT guy.

If we take this scenario seriously, there might be ethical consequences. For example, we might wish humans to exist in the physical world and not just in so many simulations. The trans-neanderthal programmer probably have the technology to embody humans, based on the simulations of what we would be like. They will only do so if we seem like the sort of creature actually worth harvesting, rather than a curiosity only worth simulating. Unfortunately, it is unclear what they might want us to do. Perhaps we should perfect our skills at making espresso, so that they might embody us to work in their coffee shops.

I for one welcome our slope-browed overlords.

[ add comment ] ( 5617 views )   |  [ 0 trackbacks ]   |  permalink
New draft on theories 
I've wanted to write this paper for quite some time, but the material from different areas has failed to cohere on the previous occasions when I've tried to write it. Now I finally have a complete draft.

What SPECIES can teach us about THEORY
ABSTRACT: This paper argues against the common, often implicit view that theories are some specific kind of thing. Instead, I argue for theory concept pluralism: There are multiple distinct theory concepts which we legitimately use in different domains and for different purposes, and we should not expect this to change. The argument goes by analogy with species concept pluralism, a familiar position in philosophy of biology. I conclude by considering some consequences for philosophy of science if theory concept pluralism is correct.


[ 1 comment ] ( 8975 views )   |  [ 0 trackbacks ]   |  permalink
Duhem? I never even... 
I am teaching Poincar and Duhem in seminar this week. They are both so sensible that reading them elicits a twinge of despair at how little progress has been made in philosophy of science since. They were ahead of their time, of course, and there have been some real advances. This is not a post about despair-- or about the bits that were so forward-looking-- so I'll move on.

In Chapter VII of The Aim and Structure of Physical Theory, Duhem claims that the modern dynamical definitions of force and power are not the meanings we get from common sense. We observe that a wagon does not move when there is no horse harnessed to it. With a horse, it moves when the horse pulls. It stops when the horse stops. Thus, Duhem suggests, common sense gives us a quasi-Aristotelean notion of force according to which force is exerted at each moment in which a thing moves. Bodies in motion tend to stop absent exertion.

The example is quaint, since we 21st-century philosophers rarely if ever see horses and wagons. The scepticism about common sense is also different than it would be for us. The 19th-century had been awash with common sense philosophy that was a kind of know-nothingism: a way for educated people to feel comfortable with their dogmas, delivered by common sense in a cellophane wrapper of critical thought.

Nevertheless, Duhem is not entirely dismissive of common sense. He offers a rather nice metaphor of common sense as a trust fund: "The fund of common sense is not a treasure buried in the soil to which no coin can ever come to be added; it is the capital of an enormous and prodigiously active association formed by the union of human minds" (p. 261). As he develops the metaphor, he suggests that anything that is appealed to as an timeless axiom of common sense is rather a withdrawl of a discovery made previously.* If the common sense claim proves unworkable, it may be replaced by a more profitable investment.

That seems right, if we conceive common sense primarily as a source of general dicta like definitions of power-- not the sort of thing we would find perfected in theories that were suited to daily life. Yet Duhem insists that common sense is right in its description of how wagons and horses typically behave. So, he concludes, "observations of common sense are certain to the extent and degree to which they deficient in detail and precision" (p. 264). They suffice as observations but fail as laws.

Since I have an axe to grind, let's distinguish three ways of understanding common sense:

1. Common sense is a reservoir of general principles: that such-and-so is the nature of power, that our senses are reliable, that all things are the work of god, and so on. This is the 19th-century, know-nothing common sense. Duhem advocates falliblism (perhaps even scepticism) about such dogmatic claims.

2. Common sense is a reservoir of particular judgements: that there is a wagon over there, that the wagons we have seen have stopped when the horse rests, and so on. Duhem calls such claims "true and certain", which seems to overstate things.

3. Common sense is a way of forming beliefs: we trust our senses unless their are specific reasons to think we are deceived, we trust our memory, and so on. Duhem does not discuss this explicitly, but it is a natural way to understand the source of the particular judgments that he thinks of as "true and certain."

And the axe that I'm grinding: Thomas Reid, that great source of the common sense tradition, is read too often as advocating 1, sometimes as advocating 2 (eg, by VanCleve), and when read correctly as advocating 3.


* William James argues similarly that what is now common sense was once a discovery. Basic truths, James suggests, were the inventions of genius cave men. The Pragmatism lectures were only a few years after Duhem's discussion appeared as articles, so I wonder if James was familiar with Duhem.

[ add comment ] ( 10466 views )   |  [ 0 trackbacks ]   |  permalink
Who put the we in the wikipedia? 
Ron alerted me to the existence of Wikipedia Scanner, a service that does the reverse lookup to follow anonymous Wikipedia edits back to their source. As one might expect, it has turned up a number cases in which corporations actively manipulated their own entries. You can get details from Wired-- or almost anywhere else, for that matter. It's already been on Colbert, making this post about two weeks behind the wave.

As someone sceptical about Wikipedia, I feel like I should have something to say about Wiki Scanner. For example, I could point out that doing reverse lookup on IP addresses only works so long as MegaHugeConglomoco does their wiki hatchet work from the corporate office. If they have it done by their PR firm or by employees telecommuting from their homes, Wiki Scanner won't make the connection. And surely any company playing Wikipedia chameleon now will make specific efforts to disguise their involvement.

All of that means that Wiki Scanner cannot generate any reliable statistics about how many organizations are manipulating Wikipedia, how many entries are manipulated, or how many manipulations remain in the present Wikipedia corpus. It sifts out anecdotes of specific abuses. Although illustrative, these anecdotes are about what one would have expected. Of course, it would be fun if it turned up some singular surprises.

Addendum Sep7


Mark alerts me to Wikipedia trust coloring, software designed to indicate trustworthiness of stretches of text in the Wikipedia. It does so by effectively making the entry history manifest in a single page. The presupposition is that text that has been in the Wikipedia through more edits is more likely to be true, or at least that users are more reliable if text they write persists through more edits.

As Mark puts it:
Now we know, in addition to <some guy said so>, that <some guy said so without other guys bothering to say not-so>.

Or better: to `some guy said so,' we add `no guy said no' (at least so said the log file); so we're better in the know.

No?
No. I mean... yes.

[ add comment ] ( 7543 views )   |  [ 0 trackbacks ]   |  permalink

<<First <Back | 64 | 65 | 66 | 67 | 68 | 69 | 70 | 71 | 72 | 73 | Next> Last>>