Monday, June 20, 2011

The External World: Preliminary Experimental Evidence of Its Existence

(collaborative with Alan Moore)

Radical solipsism is the view that nothing exists beyond my own stream of conscious experience. I’ve been wondering: Can I show radical solipsism to be false? (Reader, if you exist, then radical solipsism must be false, but – and I hope you’ll forgive me – it doesn’t follow that I can justifiably reject radical solipsism, nor that you can, with the pronoun re-indexed to your own case.)

The two most famous attempts to prove the existence of the external world are Descartes’ and G.E. Moore’s. Descartes’ crucial move was his inference from the existence of the idea of divine perfection to the conclusion that only God could be the source of such a fabulous idea – a highly dubious inference. G.E. Moore argues from “here is one hand” and “here is another” to the conclusion that “there are external things”, but he doesn’t attempt to prove “here is one hand”. Therefore, Moore starts (as he acknowledges) by taking for granted that at least one thing exists beyond his stream of conscious experience.

Thus, neither author convincingly proves the falsity of radical solipsism. Now maybe I shouldn’t need any proof that the external world exists. Perhaps, indeed, as Wittgenstein suggests, it’s a kind of philosophical sickness to want proof. So be it; I’m sick. Still, can I get what my sick mind wants?

Accepting the task at face value, the most promising approach, I think, is abductive inference or inference to the best explanation: The best explanation of why I have this pattern of conscious experiences is that something exists behind my experience that helps shape it. Maybe that’s reason enough to reject radical solipsism, even if it isn’t strict proof. But: Is the existence of some sort of external world really the best explanation of my stream of experience? I’d prefer not to wave my hands vaguely here. I wonder if I can design an experiment or series of experiments to put solipsism to scientific test.

A fair game needs ground rules. The skeptic’s position is obviously unassailable if I must prove every premise of any potential argument. On the other hand, I shouldn’t, like G.E. Moore, invoke premises that already assume the falsity of radical solipsism. For purposes of this study, then, I’ll allow myself tentatively to accept the following:

• introspective knowledge of sensory experience and other happenings in the stream of experience,
• memories of past experiences from the time of the beginning of the experiment (but not before),
• concepts and categories arrived at I-know-not-how and shorn of any presumption of real grounding in the external world,
• the general tools of reason and scientific evaluation to the extent those don’t build in any assumptions about the things beyond the stream of experience.
Accepting these ground rules, most patterns in my stream of experience don’t, I think, demand explanation by anything beyond my stream of experience. Here’s one kind of pattern: When I have an experience as of deciding to close my eyes, I experience a certain sort of dark field (or colorful Eigenlicht). It needn’t follow that I have real external eyes or am being affected by real changes in visual input. This pattern might simply be a pattern within my stream of experience, reflecting some sort of direct causal relationship between one type of experience (as of eyes closing) and another (as of a dark field). The same holds for the patterns I notice as I seem to shift my eyes around or as I seem to look at an object and then grasp it.

Nor does the randomness or uncontrollability of experience demand explanation by appeal to external things. I might have an experience as of tossing a six-sided die, accompanied by a desire for a six and a feeling of not knowing what the outcome will be, followed by an experience as of seeing a four. I might have the experience of opening a closet and feeling surprised by the contents. Such experiences might simply reflect randomness, unpredictability, and uncontrollability within my stream of experience rather than the shaping of experience by forces beyond.

What evidence, then, might I raise against radical solipsism? Here’s my first experiment. I find the experiment somewhat convincing, though I don’t find it entirely convincing – nor do I expect that you will (even if you replicate it yourself, and assuming you exist). As often in the sciences, follow-up experiments will be required to help undermine alternative possible explanations of the results.

First, I do something that I experience as like opening Microsoft Excel. Next, I do something that seems like programming Excel to generate random numbers between 1000 and 3999, excluding numbers divisible by 2 or 5. Next, I do something that seems like programming Excel to determine whether these numbers are prime, but without actually dragging down the formula to execute the calculation. Next, I briefly examine each number, one by one, and try to guess whether it’s prime. To conclude the experiment, here’s what I will do next (though I haven’t yet done it): I will do what seems like executing the calculation in Excel, then compare my guesses with the apparent Excel results, and then confirm the apparent Excel results by hand calculation.

On the assumption that radical solipsism is true, it’s reasonable to predict that either my guesses will match the apparent Excel outputs or my hand calculations will fail to confirm the accuracy of the apparent Excel outputs. I make this prediction because, if radical solipsism is true and the only thing that exists in the universe is my own stream of conscious experience, there will be no device whose calculating capacity outruns my own calculating capacity (except insofar as that calculating capacity is implicit in the laws governing direct relationships among my experiences – an important caveat that I will address in a minute).

I have generated 20 four-digit numbers as described above. Among those 20 numbers, there are 5 that I have guessed are prime. I then execute the apparent spreadsheet calculation. Excel appears to tell me that 4 of the 20 are prime, only one of which is among the 5 that I had guessed to be prime. Finally, I calculate by hand and confirm the apparent Excel results in every case, using my own mathematical reasoning. Thus, I now know that 13/20 of my guesses were correct, compared to 20/20 from the apparent Excel output. This difference in ratio is statistically significant by Fisher’s exact test (hand calculated), with a one-tailed probability of p = .004. Therefore, I conclude that there exists something in the universe with an ability to mark numbers as prime vs. nonprime that exceeds my own ability to do so, at least insofar as I am constituted wholly by my stream of conscious experience. Radical solipsism appears to be scientifically disconfirmed.

Two caveats:

First, I don’t know what that external thing is. It appears to be a Microsoft Excel program implemented on a laptop computer, but I don’t regard myself as having established that fact. The calculation might even have been executed by a nonconscious part of my own mind, in which case some weak form of solipsism might still be true.

Second, there are, of course – as there always are in scientific research – possible alternative explanations of the results. The ground rules of the current exercise require that I consider these explanations using ordinary scientific standards insofar as those standards can be implemented while bracketing the question of whether an external world exists. Some alternative explanations I hope to rule out in future experiments. But at least one seems worth mentioning right away: I have already granted that there may be laws that relate experiences to each other without implying the existence of anything beyond experience. These laws might in fact have a level of organization beyond what I can consciously appreciate as they unfold. For example, the laws relating the colors of apparent visual stimuli with the complementary colors of the afterimages produced by those apparent visual stimuli may involve regularities that I can’t immediately appreciate. Could there, then, be some similar law of experience relating the four-digit numbers in the relevant Excel column with the marking as prime vs. nonprime in another column?

That would seem a strange law for the following reason: The image/afterimage law and other similar laws or patterns of experience, such as the eyes-open/eyes-closed pattern and patterns of relationship between visual appearance and tactile appearance, are laws that relate surface features of one experience with surface features of another, e.g., color of image to color of afterimage, smoothness of visual appearance to smoothness of tactile appearance. In contrast, the proposed experiential law governing my apparent Excel outputs concerns a non-superficial feature of my visual experience – the number represented by the numeral I interpret on the basis of the apparent screen pixels. The posited solipsistic law would relate semantic features of one visual experience to semantic features of another, with the first set of features apparently undetectable by me (and thus by any person, if solipsism is true), though the features were subsequently confirmable by me through laborious calculation. To posit the existence of such a weird semantic law relating two sensory experiences of apparent Excel columns seems like an ad hoc maneuver to save the solipsistic theory, lacking independent motivation, and therefore dubious on scientific grounds.


gwern said...

> Second, there are, of course – as there always are in scientific research – possible alternative explanations of the results.

There are, and far more plausible ones than 'disproving radical solipsism'.

For example, you could simply be a savant - factoring numbers is one of the standard tricks (although see ). Savant-like abilities are even inducible to some degree by transcranial magnetic stimulation (

Theoretical estimates of the computing power of your brain come in at many orders of magnitude more power than your computer running Excel (or indeed, how much computing power it takes to factor small numbers like those).

Even if you are willing to grant the would-be disprover of solipsism the *entire* edifice of modern mathematics and complexity theory, you still don't prove anything with the factoring example, because factoring has not been proven to be harder than P. (What exactly it is is a huge open question and the bounds are a mess: ).

These general points apply to any attempt to draw upon resources in the 'outside world' - often, you have outside-world evidence that you *can* do the task yourself, and even if you don't have such evidence, where do you get your proof that you don't have hidden resources to draw upon?

Eric Schwitzgebel said...

Gwern: Good points, but let me call your attention to my first caveat. If the computing ability comes from an unconscious savant-like ability or from from hidden resources in my brain (which I cannot see and which the radical solipsist doesn't know exists), then radical solipsism is still false, though a weaker form of solipsism might be true.

gwern said...

Hm, I think I see. So by 'radical solipsism' you mean everything must be in consciousness, that even accessing a previously unknown part of your own mind (like a savant ability or forgotten fast factoring algorithm) counts?

But then, there's a lot of stuff that comes into consciousness that was not there before, like random memories or thoughts. Often you recall something you didn't know that you knew. Should we take all these mental transients as disproving radical solipsism too?

Switch said...

It seems to me that even radical solipsism admits that the stream of consciousness is outside of the solipsist's direct control. Even had you the ability to correctly calculate whether the numbers were prime or not (as Excell did), controlling when that occurred may have been out of your capabilities, making this series of events possible even if all that exists is your stream of consciousness.

Eric Schwitzgebel said...

Gwern: Things might leap to mind unbidden without providing evidence against solipsism (that's like the case of rolling a 4). And even a certain kind of memory is probably okay: If I see a certain visual display, close my eyes, and then see it again, the relationship between the two *might* be just a relationship between the pre-eyes-closed and post-eyes-closed experience, without an intervening non-conscious medium. This might seem to require action at a temporal distance, which is a bit funny but in the context of solipsism I think it's reasonable to spot the solipsist that point.

Eric Schwitzgebel said...

Switch: It seems to me the present case is not simply a beyond-my-control case (like the die-rolling case or the closet case). I believe it is a bit harder for the solipsist to allow for computation beyond my ken than to allow for mere lack of control and surprise. You've put your finger on the heart of the matter, though, I think.

John Dell (administrator) said...

Geez, Eric!? I'm going to leave the details of your argument alone...though I'm not sure the solipsist needs to grant that you be "aware" of your mathematical prowess; surely your powers of introspection are fallible, no?. So wouldn't the solipsist just continue to claim that you have no way of knowing what's "outside your ken" and what isn't? Maybe this Excel you speak of is just a way for you have an accelerated (and more accurate)experience? After all, there is no requirement that you know the *origin* of your experience...

My concern is whether solipsism could ever be proved wrong scientifically. The problem here, as I see it, is that for something to be truly scientific you need repeatability from other researchers, but if you are trying to prove that there is an external world, and to do that you need repeated measures from an external world you do not know exists, you can only appeal to repeatability and truly invoke the scientific method if you assume what you are trying to prove. In other words, repeatability can not be used to show solipsism is incorrect because you have no way of knowing whether there is anyone else *actually* doing the repeating. Of course, there is also the fact that scientifically speaking you never really "prove" anything. So really I don't see how you ever get away from the "inferences" that you seemed to shy away from at the beginning of your post.

To be honest, I think Wittgenstein hit it on the head with this one. Still, my favorite way to treat a hypothetical solipsist (S) is horribly brute. I would like to set S up against a wall and prepare to throw things at him. I would first make sure S consents that there is no reason to duck since these things I have gathered are not here. Then I would begin by throwing a Nerf ball. I would imagine S would take this in the face with little difficulty, but by the time I got to the hammer, I imagine S would be ducking rather quickly (especially if we ever got to the second throw with the hammer). The best I imagine S could say here is that he sees and feels a hammer, but does not believe it is there. But this sounds to me like saying, "it is raining, but I do not believe it". So with brute force can you throw the solipsist into Moore's paradox? And if you can do that, then can't you move to throw doubt on S's belief in solipsism?

I know this is horribly crude, and I'm not arguing to prove solipsism false, I'm just arguing that there are empirical ways to beat the solipsism out of a person that might believe it (a strange therapy indeed). Which for me raises another question: if no one believes in solipsism is there any reason to worry about it?

Eric Schwitzgebel said...

Ouch! *ducking out of the way*

One thing I need to get clearer about is whether this is a valuable exercise given that, as you say, no one believes solipsism.

Jeff M. said...

"Things might leap to mind unbidden without providing evidence against solipsism"

Things might leap to mind unbidden... from where exactly?

John Dell (administrator) said...

Yes, it's mot an easy question. For the last few years I have generally sided with Peirce on such matters. His position was more or less that it was okay to engage in such intellectual pursuits as long as you emerge from your den of thought with something worthwhile to the way we live our lives and understand the world.

Badda Being said...

>First, I do something that I experience as like opening Microsoft Excel.

Can you explain what you mean by the word 'like' in this and subsequent sentences in the same paragraph? It seems like (or as if or by comparison that) you are invoking a relation of similitude between two things that you already assume to be real, namely, the experience of opening Micorosoft Excel and actually opening Microsoft Excel.

Otherwise, on the one hand, if you are not here drawing a distinction between experience and actuality, then your statement seems tautological along the lines of saying: I do something that I experience as like the experience of opening Microsoft Excel. On the other hand, if your use of 'like' here does not have a relational function, then your statement is incomprehensible to me.

Eric Schwitzgebel said...

Thanks for the continuing comments, folks!

@ Jeff: E.g., arising randomly from the stream of experience, in accordance with some pattern. Compare randomness in quantum mechanics, e.g., the spontaneous creation of particle/antiparticle pairs. (Agreed, it's a bit of a strain; but I don't think it's a defeater for solipsism.)

@ John: I confess to some sympathy with pragmatism. On the pragmatic value of skepticism, see Sextus Empiricus.

@ Badda: I meant the "like" to convey that I had a certain experience that seems to be an experience of opening Excel, but neutral regarding whether I really did open an external Excel. Perhaps there's a better way to put it?

Badda Being said...

I suspect there is no better way to put it that would validate the terms of your experiment. In this alternative construction you are a justification short of being in violation of your second ground rule. But then there is no sensible way in which that statement could have been formulated within the confines of the experiment. On what basis did your experience seem to be an experience of opening Excel? A basis can only found from a transcendental viewpoint relative to the experiment, adducing either memories of past experience prior to the experiment or the reality external to what you experience within it. So, ironically, this experiment can only be performed in your head while sitting in your armchair.

Badda Being said...

No, I get it. The seeming of an experience -- a meta-experience, an experience at a higher level of exponentiation -- can still be neutral with respect to the exteriority of its base.

Badda Being said...

In my haste to dismiss your experiment I had previously overlooked your third ground rule.

And yet, and yet: if ad hocness is sufficient for rendering a proposition dubious on scientific grounds, then radical solipsism is already dubious without the intervention of your experiment, yes no? Admitting one ad hoc proposition (radical solipsism) and shirking from another (a semantic law) seems like an ad hoc maneuver in and of itself to get the experiment off the ground and steer yourself towards a particular interpretation of the results.

Still, I like this, I like this.

G. Randolph Mayes said...

I f-ing love this post.

G. Randolph Mayes said...
This comment has been removed by the author.