Wednesday, September 14, 2011

The External World: More Experimental Evidence of Its Existence

(collaborative with Alan Moore)

I’m engaged in a certain exercise. You might doubt the value of this exercise, but it’s a traditional philosophical exercise, and I’m a philosopher. The exercise is attempting to prove the existence of an external world beyond my own stream of experience.

As I mentioned in a previous post, the two most historically famous proofs of the external world appear to be unsatisfactory: Descartes’s because it turns on the dubious claim that the idea of perfection could only be caused by a perfect being, and G.E. Moore’s because it starts from the question-begging premise “here is a hand”.

My strategy is this: First, I assume introspective knowledge of my current sensory and other experiences, memory of past experiences back to the beginning of the exercise, and the general concepts and methods of science insofar as those are stripped of any presupposition of a world beyond my stream of experience. Then, I attempt to find experimental evidence sufficient to justify belief in a world beyond my stream of experience, as the best scientific explanation of patterns in my stream of experience.

In Experiment 1, I did something that seemed like programming a spreadsheet to determine whether various four-digit numbers were prime, and I also guessed whether those numbers were prime. As confirmed by subsequent hand calculation, the seeming spreadsheet did a far better job than I did at correctly marking the primes, and the best explanation appeared to be that something exists with calculating abilities exceeding my own – at least insofar as the “I” is conceived solipsistically, as constituted entirely by my stream of experience.

Experiments 2a and 2b:

Now I will try a second experiment, using an apparent confederate. I will call this apparent confederate “Alan”, though without meaning to presuppose that he exists as anything but a figment of my own stream of experience.

This “Alan”, by what I think of as prearranged instructions, gives me a list of 20 three-letter combinations to memorize (“EMA”, “GLL”, etc.). The list of three-letter strings appears orally, by which I mean that I seem to hear him say the 20 three-letter combinations. He appears to present the list twice. I then attempt to freely recall the letter-combinations from the list, offering up six guesses – the three-letter strings that it seems I can recall without prompting. “Alan” then presents me visually with a forced-choice recognition test: 40 three-letter combinations, half of which he says are new and half of which he says are the original 20, in random order. I select 20 of the 40 as my best guess as to which are the old ones. I feel high confidence about 10 of these guesses. We then repeat the same experiment with a second set of three-letter combinations. This time I offer up eight guesses in free recall and in the 40-item recognition test I again feel high confidence about 10 items.

Now, if nothing exists beyond my own stream of experience, then the forgotten items on these lists should not, it seems, be stable. They exited my mind – either as measured by free recall or by the stricter measure of recognition – and thus they exited the universe. Thus, when Alan seems to present me with these lists a second time, which he will shortly do, nothing endures across time that could anchor the forgotten items in place, ensuring that the new perfectly matches the old. The seemingly re-presented items could be any set of plausible items; my mind has unconstrained liberty to invent any letter combinations that seem reasonable candidates to span the gaps in my memory. On solipsism, the present needn’t preserve the details of the past; it need only preserve, seemingly can only preserve, the details of the past that I remember.

However, Alan now shows me evidence that undercuts this solipsist picture – evidence suggesting that the 20 three-letter combinations that were the “right” answers in each test were in fact anchored in place and stable across time rather than created at liberty anew. He shows me, first, the lyrics of “Take Me Out to the Ball Game”, and he tells me that the first 20 three-letter combinations were the letters, in reverse order, from the end of the chorus of that song (excluding spaces, punctuation, and two repeated letter combinations). As I look through that list of letter combinations (“EMA”, “GLL”, “ABD”, etc.), it is quite evident to me that this is the case: I am forcefully struck by the fact that the letter combinations that I do seem to recall from the original list fit exactly into that pattern, though I was entirely unaware of that pattern at the time they were originally presented. Comparing my free recall and recognition lists with the reversed-ballgame song, I see that 5 of my 6 freely recalled letter combinations appear in that song (one guess, Alan tells me, was wrong) and that 15 of the 20 I seemed to recognize were correct, including all 10 about which I felt confident in the recognition test. The other stimulus, Alan tells me, was similarly derived in backwards chunks from a sentence from George Berkeley: “In short, if there were external bodies, it is impossible we should ever come to know it; and if there were not, we might have the very same reasons to think there were that we have now.” Here, Alan tells me I accurately freely recalled 7 of the 20 combinations (having made one erroneous guess out of eight) and correctly recognized 16 out of 20, including all 10 about which I had felt confident.

Why should I believe that these lists now shown to me by this seeming Alan actually match the original ones presented? Their structure suggests so: Although I am not assuming the existence of ballgames or peanuts or cracker jack, it is one of the background assumptions of this exercise that I permit myself my existing conceptualizations, stripped of any commitment to a really existing external world. And my conceptual structure contains the song “Take Me Out to the Ball Game”, in which these words appear as indicated. Likewise, it’s part of my conceptualization of George Berkeley that this is just the sort of thing he would say and of Alan’s sense of humor that he would use this sentence to generate the stimulus. While it’s remotely possible that from the fragments that I happened to remember such natural-seeming structures could be found post-hoc to fit, that would seem to require enormous happenstance. Suppose that there were as many as a trillion possible strings of text Alan could have used, per the instructions I seem to remember having given him, to generate suitable test material, given present purposes. And suppose conservatively that English allows only about 1000 viable, roughly equiprobable, plausible three-letter combinations (it would be about 18,000 if all 26 letters could appear in any combination). Still there would be only about a one in ten billion chance that I could find two matches from that trillion strings for my two sets of ten confidently recognized three-letter combinations.

A better explanation, I venture, than such a boggling chance is that these remembered letter combinations were indeed generated in something like the manner described by this “Alan”, and thus that the lists Alan is now showing me do, as he says, match the original, in both their remembered and their unremembered parts. In other words, this pattern among my remembrances was not constructed anew but was present from the beginning in the original experience of hearing the letter combinations, including in the unremembered parts of that experience, since without those unremembered parts the strings would not fit the stated pattern. But then these lists, or at least the pattern they reflect, must have endured beyond my conscious ken from the time of original experience to the time at which the pattern was revealed to me, sustained by some mechanism running outside my stream of conscious experience.

Maybe my knowledge of the ballgame song or the Berkeley sentence operated nonconsciously in me to generate those letter combinations from the start? But that hypothesis still involves rejecting the most radical form of solipsism, the form of solipsism currently under discussion, which posits that all that exists in the universe is my own stream of experience. I don’t conclude that I know what, exactly, is beyond my experience – that Alan exists or that the physical list exists – only that something is beyond. That something might be my nonconscious mind.

Maybe my conceptualization of song lyrics and great philosophers is, unbeknownst to me, highly labile, so that although it seems that the lyrics are annoyingly familiar and the quote from Berkeley oh-so-apt, in fact this sense of familiarity and aptness arose wholly for the occasion, via some strange law of relating past and present experience, and the combinations are actually new, perhaps even the words themselves new English words? Then there need have been nothing outside my own conscious experience that endured from the first moment of presentation until now. This objection seems ad hoc given the background rules of this exercise, but in any case, it can be avoided with a third experiment.

Experiment 3:

Experiment 3 is structured much like experiments 2a and 2b, except that the items to be remembered are 20 three-digit numbers. So: I (and “Alan”) run the experiment. Alan tells me, now, that I have correctly freely recalled eight three-number combinations (nine guesses with one error) and have correctly recognized 15 items, including all of the eight combinations about which I felt confident during the recognition test. What suggests the stability of the list in this case? Alan tells me that the 20 three-digit numbers are actually the first 60 decimal places of the long cyclic number that comes from dividing 1 by 1939, the year in which G.E. Moore delivered his famous “Proof of an External World”. This time, instead of trusting in the stability of my conceptualization of song lyrics, I plan to trust to the stability of mathematics, confirming the list’s stability, the existence of the pattern Alan alleges to unite the initial and present stimuli, by long division by hand to 60 decimal places.

But, alas! After the first six digits something has gone wrong. Hand division does not confirm the list; the numbers diverge. Have I made a calculation mistake? No, evidently not, unless I am very deeply confused. Is there, perhaps, no external world or stability to mathematics after all? I suppose I am a biased experimentalist; I am reluctant to draw that conclusion. The number is a bit larger in the 7th digit, so on a hunch, I check the decimal expansion of 1/1938. Indeed, there is a perfect match all the way out to the 60th digit, between the list Alan tells me was the original stimulus and the hand-proven results of my calculation. (N.B.: In subsequent conversation “Alan” confirmed the diagnosis, but that small slice of experience is outside the scope of the experiment and thus out of bounds for now.)

I seem to recall having suggested to Alan that he use a long cyclic decimal expansion or some other readily calculated but unpredictable number string to generate the list for this experiment. Supposing that there were a million suitable pseudo-random number strings from which Alan might have chosen, the odds are over a million to one that all eight stably remembered three-digit combinations would appear among the 20 sets of three digits from one of those million candidate number strings. Again, the more likely explanation appears to be that the eight remembered combinations were indeed generated by the discovered pattern, along with the other unremembered combinations, and thus that the present list matches the original list and consequently that something endured over time, preserving that structure outside my conscious ken, contra the radically solipsistic hypothesis. This conclusion appears justified even though it took some searching to find the driving pattern.


The resolute skeptic will, of course, find joints at which to challenge this argument: Maybe the laws of math aren’t stable over time. Maybe there’s unmediated action at a temporal distance between my first oral experience of the list and my subsequent visual experience of the list. Maybe my memory is so poor that I can’t even trust that I did in fact recall several items correctly and this whole business has been an illusion of the past five seconds. In general, anywhere there is an undefended premise, the skeptic can insert a knife; and all arguments must have undefended premises on pain of regress. But recall the aim of this exercise: It is to provide experimental evidence, in the form not of incontestable proof but rather of inference to the best explanation, for the existence of something beyond my stream of experience, while taking for granted knowledge of my stream of experience and my pre-existing scientific standards and conceptualizations except insofar as those standards and conceptualizations presuppose the existence of anything beyond my stream of experience. That, I tentatively believe I have done.

Proposals for a fourth experiment warmly welcomed.


Neil F said...

I think we can boil this down:

There's a chunk of information S which is in some way 'special'. (Ideally the meaning of 'special' should be fixed at the outset, in such a way that 'specialness' is rare.) You're given a little subchunk s of S. Later you're given the whole of S (= s + t, let's say), and you marvel at how unlikely it is that your mind could have spontaneously generated an s for which a t exists such that s + t is 'special'.

You reason that it's much more likely that t existed all along, somewhere external to your conscious thoughts.

Anonymous said...

I like this approach because it doesn’t attempt to go via the accuracy of what seem to be our perceptions. You could alter the experiments so that even a being who is situated in some world external to its stream of experience, but who is having some strange hallucination rather than perceiving that world, could show to its own satisfaction that there is something external to it. E.g., as long as in its hallucination there is some reliable means of quickly identifying large primes as primes that surpasses its conscious ability to identify primes (assuming that in the hallucination it’s lucid enough to run the experiment), it can think to itself that there is probably an external explanation for this.

Of course, once we come to believe there is an external world, we’ll want to learn about it. You’ve been careful not to get into that, of course, but once your result as to the existence of an external world is published, I think the exultant scientific community will demand research into whether we can have knowledge of the external world in the way we appear to.

Where would you have this published, by the way? In the leading journal of some field, surely; probably physics - since you’ve proved, granted some plausible assumptions, there is at least one body - or chemistry - since you’ve proved, granted etc., there is at least one substance. I think economics journals might be interested too, though. After all, there’s no fun in game theory or in trade if only one person exists. But you can show that there must be at least one other, because it can appear to you that you are facing some other person in a number of games of chess, and that they consistently beat you with moves that you can’t see to be dangerous at the time, but that you can later analyse to be tactically solid rather than merely lucky. This is much like your prime number experiment (if you were conscious of all that existed, what would explain the reliable effectiveness of moves that you can’t see to be effective at the time?) and makes it plausible that there is at least one other chess-playing agent, even if they don’t have all the other qualities of the person you appear to see on the other side of the board. You might want to put some numbers in before you send this to a journal of physics or economics, though. Can you get P(There is an external world) > 0.95?

Eric Schwitzgebel said...

Neil F: That's an interesting way of boiling it down! Thank you for the suggestion. Alan and I will have to think a bit more about whether this really is the nut of the argument, or whether it's only close. In any case, it's a helpful wedge into thinking about what the several experiments all have common.

Eric Schwitzgebel said...

Anon Sep 15: Thanks for your kind remarks -- too kind, I suspect! I'm inclined to think that physics and economics journals wouldn't have much interest in this. We'll be pitching it as "experimental philosophy" when and if we decide it's ready for formal publication.

As for what the external world is like, in terms of basic metaphysics and cosmology I currently favor what in an earlier post I called "crazyism".

Marshall said...

You're assuming you have complete knowledge of the contents of your memory, which is clearly not the case for humans. On the contrary; in a simple case: you can't find your keys; only when you locate them on an unusual shelf do you remember putting them there. Not to mention tricks of avoidance where the mind "deliberately" hides relevant facts from consciousness. You want to admit your subconscious as part of your self, don't you? Or do you intend these rationalist beings as a thought experiment?

I think perhaps you are making a clean break between the stream of events that is your experience (which is completely known) and that larger stream of events we call Reality (which is beyond personal control). I'm not sure you can do that, considering that the former is an interior subset of the latter, but maybe.

For another (natural) experiment ... I would suggest considering the experience of pain and suffering. At times pain can be gratifying, but let's call suffering an experience of pain which is intolerable although unavoidable. If you could relieve your suffering by an act of will, then surely you would, so suffering is apparently imposed by something beyond your will. This will not be convincing to people who aren't or haven't suffered, but why would people like that ask a question like this at all?

Anonymous said...


I may be wrong, but I think Professor Schwitzgebel's exercise does not assume the ability to actually control his stream of experience in any way. So suffering would not constitute evidence of the existence of the external world.

Eric Schwitzgebel said...

Yes, anon: I'm assuming that lack of control of my experience is not by itself evidence that there's anything beyond my experience. And, Marshall, I am also making certain assumptions about the quality of my memory. Arguably, one could go the opposite direction on both assumptions and end up with a different exercise. But for what it's worth those are part of the ground rules of the current project!

awesomesauce said...

This is ridiculously interesting!

Jackson Davis said...

According to Orteguian metaphysics, "ex-peri-ments" only occur in your "ex-peri-ence" which is "ex-ternal" to "you" but "in-ternal" to "your life".

Also for your "ex-peri-ment" to be "scientific" you must set up a negative hypothesis that there is no correlation between your recalled auditory or visual experiences and the subsequent observed or reported experiences. You would then attempt to prove or disprove that negative hypothesis. Science never "proves" a hypothesis. The best it can do is to fail to disprove the corresponding negative hypothesis. In the case of a correlation among multiple events ("experiences") of the kind you describe, the correspondence as you indicate is statistical. Therefore you would have to phrase your negative hypothesis in terms of "There is no statistically significant correlation between your remembered experiences and the subsequent observed or reported experiences." You would then have to make a judgement as to what would be "statistically significant".

I'm not familiar enough with the methodology of so-called "philosophical experiments" to be able to evaluate whether or not your "ex-peri-ments" would qualify.

Eric Schwitzgebel said...

Jackson: Thanks for your comment! Obviously this isn't the usual sort of experiment, so the standards are a bit different, as you notice. I'm not sure I accept the background philosophy of science in your comment, however. Obviously that's a big issue!

Jackson Davis said...

There is no "philosophy of science" involved. This is the way that science operates on a day to day basis.

Web Monster said...

I believe Gödel's incompleteness theorems seriously weaken your arguments about mathematical stability, and it does not seem as if you investigated the philosophy of mathematics before calling it 'stable'.

Eric Schwitzgebel said...

Thanks for the comment Monster. The background assumptions are supposed to be those of ordinary science, bracketing the existence of an external world. In ordinary scientific usage Goedel's proof is not ordinarily taken to call into doubt scientific reliance on the stability of 4-digit primes and decimal expansions of 4-digit fractions. Yes?

Stan G said...

What if you tried to find a line not between your stream of experience and the external world (if it exists at all), but a line between you yourself and your stream of experience instead. What if you found that your stream of experience is already external - memories, rules, scientific principles etc?

I don't think it would directly help your radical solipsism problem but it would surely put it in a different light - is there anything external to the externality you declare as "yours"? How much of the external world you lay claim to? All of it or just a part?

Also, if you don't have control over what happens in your stream of experience, then who does? If your stream of experience is independent of you, does it depend on anything else, perhaps external to it? Or does it have life of its own? How would that work?

Maybe it would be easier to deal with the narrow problem of radical solipsism if you put it in the broader perspective. Right now all kinds of things pop up unexpectedly - insufficient probability, knowledge of the songs or mathematical divisions and what not. If we knew where these things belong in the first place it would be much easier to avoid them when framing your experiments.

Eric Schwitzgebel said...

Another interesting, and tricky, set of questions, Stan! To tricky, though, and contentious to solve, for me to want to incorporate answers to them into the present project.

For what it's worth, I am pretty skeptical about robust and determinate conceptions of the "self"; my perspective on that is broadly Humean, Parfitian, Dennettian.