Blog Coordinator

Knobe's X-Phi Page

X-Phi Grad Programs

« Experimental Philosophy Lab Meetings in New York: Bateman and Buckwalter and Devitt on 2/11 | Main | How the Experience Machine Works »



Feed You can follow this conversation by subscribing to the comment feed for this post.


Great point! Just two days ago I was wondering whether some experiments I've been running (and other potential future experiments) would fall under the positive or negative project, and I could not really justify placing it within either camp.

I do like this distinction better, but might we be able to consider *both* ends of the distinction in a way similar to the a priori/a posteriori and analytic/synthetic divides, where you end up four potential categories? After all, it seems to me that just as one can understand the first of the new distinctive methods of X-phi as being either a positive or negative project with regard to the traditional views in analytic philosophy (just as you pointed out), likewise, one might understand the latter method of X-phi as having a positive/negative distinction in that some seek to shed light on human nature and provide explanations of how and why human minds function, whereas another sort of project might be to use the second method to help do away with any conceptions of human nature and rather to illuminate certain discordant cognitive processes which might question the overall view of what it is to be human - or a self/decision-maker - in the first place.

This is a crude way of putting it, surely, but I hope I got my point across. Maybe in the spirit of the analytic/synthetic / a priori/a posteriori table mentioned above, is it possible that there are actually (at least) four kinds of projects one might be doing in X-phi, namely, the two new distinctions you've listed above, but with a positive and negative approach to each?

jonathan weinberg

I had a longish, complicated comment worked up, but then on reflection it seemed to me that this isn't complicated at all: you've just recapitulated part of the already-standard taxonomy. Remember, within the positive program there generally taken to be two main sub-flavors, which Eddy & Thomas called "Experimental Analysis" and "Experimental Descriptivism". The latter _just is_ your option (b). And your option (a) is just the union of the "Experimental Analysis" stuff and all the negative program/restrictionist stuff.

It's useful to observe, as you have, that there's a similarity to take note of between the "Experimental Analysis" and restrictionist projects. But that doesn't really seem to me to make any trouble at all for the existing taxonomy. One could also group the three main projects in a way that puts restrictionism and descriptionism in one box, and analysis in the other: restrictionism and descriptionism do not look to make substantial use of intuitions (armchair or otherwise) in any direct way as evidence for traditional philosophical questions, whereas the analysis approach substantially endorses such uses (but not from the armchair). Though I hasten to add: I generally try to avoid theorizing about x-phi on the whole in a way that builds anything about intuitions right into the very fabric of x-phi.

(Btw, I don't see how there's any question about how to taxonomize the moral/conventional work -- it seems to me completely of a piece with the descriptivist approach, and shares very little (except some key authors!) with the restrictionist program.)

Having said all that, I don't think that the existing taxonomy gives us a seamless partition of x-phi space. What's somewhat harder to taxonomize, I think, are people who want to use the deeper psychological results not just for their own intrinsic philosophical value in a human-nature project, but to shed light back on some of those "issues in analytic philosophy". I take Eric S's work to be of this sort, for example; also I would include here arguments by various x-philes (including some of your work with Shaun) that attempt to reveal which of a set of conflicting intuitions ought be trusted, and which rejected. But it seems to me that these don't fit well in your own proposal for a revised taxonomy, either.

Timothy Scriven

Interesting post. I suspect you simply want too much from the distinction. As with most academic fields most of the time, no subdivisions hold together if you push them.

I think that there are papers which tend to try to solve or contribute to the solutions of the traditional problems of philosophy, using intuitions to do so, but in an unusual way. There are also papers which try to challenge intuition as a source of evidence in these domains. Yes there are many confusing cases, but a distinction needn't be perfect, it just has to do work.

Joshua Knobe

Thanks for these helpful comments. I feel like we are definitely converging in some ways here, and maybe we will soon be able to arrive at even more of a consensus.

The main point of my original post was about the change in experimental work over time. If you go back to the very early days of experimental philosophy, it definitely was helpful to distinguish in this way between 'positive' and 'negative' projects. There were a substantial number of experiments that fit pretty clearly into each project, and although the distinction wasn't perfectly clear or exhaustive, it definitely did provide a relatively good description of what was going on. (Even now, when people try to characterize the two projects, they almost always do so by reference to these early experiments.)

My sense, however, is that this schema doesn't really capture what has been going on in, say, the past five years. Just as Jonathan mentions at the end of his comment above, these recent years have seen the emergence of a profusion of different approaches that don't fit too comfortably into the earlier categories: Schwitzgebel & Rust study the intuitions of moral philosophers and compare them to these philosophers' actual behaviors; Buckwalter & Stich look at gender differences in intuition and use these differences to explain the gender gap in our profession; Sripada argues that people understand human psychology in terms of a 'deep self' but does not take any stance on whether this deep self actually exists; Machery now argues that cultural differences provide evidence for a 'relativist' theory of reference; much of the recent work in epistemology shows that earlier epistemologists were wrong in their claims about the impact of stakes but does not clearly take either a negative or positive stand; Nichols, Nahmias and many others have developed research programs that involve arguing that different intuitions arise from different psychological processes and that a careful look at these processes will help us figure out which intuition is most deserving of our trust.

Keeping in mind that there have been hundreds of experiments in these past five years or so, my sense is that it would not be possible to characterize some substantial percentage of the total as 'positive' and some other substantial percentage as 'negative.' Instead, I was thinking that if we just looked at these experiments -- and didn't come in with categories already in hand based on the earlier research -- we would probably have divided them up in some very different way.

Of course, some researchers might disagree with me on this point, but I wonder if they would at least agree to the compromise position that the distinction is not as helpful in characterizing recent experimental work as it is in characterizing the work that was done in the early years of experimental philosophy.

jonathan weinberg

I dunno, Josh, maybe we just disagree about how useful the distinction was 10 years ago! It seems to me to be pretty much just as useful now as it was then: it's a rough heuristic, highly useful for distinguishing the general shape & direction of various families of projects and kinds of arguments, but obviously no substitute for close attention to the particular arguments themselves. This is especially so when we keep in mind the clear sub-division within the "positive" side (at least in some part the problem here might be the result your small, but nontrivial, mischaracterization of the distinction in the first place, which collapsed that subdistinction.)

Of the examples you listed above, most are very straightforward to categorize within this framework, with S&R, Sripada, and B&S would all falling straightforwardly into the 'descriptivist' category; and the epistemology stuff you're referencing pretty clearly, it seems to me, on the 'analysis' side. (The descriptivism/analysis distinction may itself break down for some projects, like that of articulating the contours of how we do or don't ascribe knowledge. But that's still fully on the 'positive' side of the main distinction.) I don't know the argument of Edouard's you're referencing, unfortunately, and I already addressed the Nichols-type arguments in my previous post.

So it still seems to me to be pretty useful, but not at all a precise partition; and that's all it was ever meant to be. No?

Joshua Knobe

Hi Jonathan,

First, let me say that I really appreciate your point about this approach to experimental philosophy (associated, e.g., with Nichols) that looks at different psychological processes and tries to get a sense of which might be more reliable and which less. That's definitely a helpful addition to the discussion here.

Anyway, my concern was that any way of dividing up the field as a whole should end up assigning a substantial percentage of experiments to each of the different categories. If you go back a few years to the time when this distinction was first introduced, there were a large number of studies that were specifically designed to contribute to something like the positive project (including, e.g., the seminal work of Nahmias and colleagues) and also a large number of studies designed specifically to contribute to a negative project (including, e.g., your amazing work on attributions of knowledge). Each of these types of studies accounted for a pretty healthy percentage of the total amount of experiments being done at the time.

But these early studies ended up serving as an inspiration to other researchers, and over the next few years, all sorts of people started pursuing all sorts of different projects. So if we just look at the work being done over the past few years, we wouldn't find, e.g., that any substantial percentage of the total number of experiments were designed specifically to advance a negative program. (Rather, what we might find is that early work in that mold had inspired these researchers, who were now doing all sorts of different things.) It is precisely in that sense that I worry that this approach to categorization doesn't quite help to characterize the nature of the field as it is right now.

Building on the suggestion you make above, though, we might be able to describe a slightly broader category that actually does account for a substantial percentage of the total. Specifically, one could say that there is a certain brand of research that looks at the psychological processes underlying people's intuitions as a way of trying to figure out whether we should trust those intuitions. This category would then include everything that was explicitly negative, but also a bunch of other work besides.

Ben Gadd

It seems as though the terms "positive" and "negative" are potentially misleading in terms of the way you've described the classification scheme, and I wonder if that may be the reason the labels seem questionable now.

Papers that do what you've indicated as the project of "negative" papers would be papers drawing explicit conclusions about the proper methodology for philosophical inquiry (a meta-philosophical project). Papers that do what you've indicated as the project of "positive" papers would be those that use experimental methods to draw conclusions about first-order philosophical debates.

I would hope that the distinction between a) metaphilosophical papers whose conclusions support experimental methods over armchair methods and b) papers which use experimental methods to support first-order philosophical positions remains useful for categorizing work in ex-phi. The "positive" and "negative" distinction seems to cross-cut the meta vs. first-order distinction. For instance, there could be:
i) A paper using experimental methods to argue against some particular popular analysis of knowledge (negative, first-order).
ii) A paper using experimental methods to argue in favor of some particular popular analysis of knowledge (positive, first-order)
iii) A paper arguing against arm-chair methodology, using evidence from experimental philosophy (negative, meta)
or iv) A paper arguing in favor of experimental methods, using evidence from experimental philosophy (positive, meta).

I don't mean to suggest that these categories are exhaustive (they probably aren't even exclusive, as papers will frequently be up to more than one thing at a time), but it is unclear why these categories would be obsolete.

Joshua Knobe


That's a nice conceptual point. Just as you say, there are really two different distinctions here (first-order vs. meta, positive vs. negative), and one could imagine any combination of these possibilities.

My one worry about this classification scheme, again, is that it doesn't help to capture what people in the field are actually doing these days. (No one actually is running experiments for the purpose of arguing positively for the idea of experimental philosophy.) So my thought is that if we want to capture the nature of the work that people are actually doing at this point, it might not be helpful to include this negative/positive division at all.

Thomas Nadelhoffer


I think the point Jonathan was trying to make before is that your distinction between positive and negative hides an important distinction that Eddy and I highlighted in our "The Past and Future of Experimental Philosophy." There we put forward the following three-part distinction:

"Despite the fact that all of the highly collaborative and interdisciplinary work that has been done by experimental philosophers so far has been driven by these overarching commitments, there are at least three distinct kinds of projects that have been undertaken. While some work in experimental philosophy seeks primarily to find out what the folk think in order to ascertain which philosophical theories best accord with and account for commonsense intuitions, other work primarily aims to explore how the folk think—what psychological mechanisms produce people’s intuitions or theories—and to determine the relevance of this information to philosophical disputes. Finally, a third project within experimental philosophy relies on data concerning cognitive diversity to argue that philosophers should not use intuitions as evidence in their theorizing."

We called these three projects or approaches to x-phi Experimental Analysis, Experimental Descriptivism, and Experimental Restrictionism (which is a term we borrowed from Jonathan and Joshua's nice piece on x-phi work in epistemology). So, while you may be right that the negative/positive distinction might sometimes be too broad to capture the current state of the field, the EA/ED/ER taxonomy Eddy and I developed seems to provide you with the ability to draw the distinctions you're after. If you don't agree, I would be interested to hear which current experiments or projects don't fit at least loosely into one (or more) of these three boxes.

Joshua Knobe

Hi Thomas,

I really like that trichotomy you offer in the article with Eddy, and I think it does a great job of characterizing the main strands of work that you were reviewing at the time. My sense, though, is that this early work ended up inspiring a whole bunch of different kinds of further experimentation, with lots of different researchers doing lots of different things. So it might still be possible to classify each of the non-descriptive papers as 'positive' or 'negative,' but I'm not sure that this classification is nearly as helpful in characterizing what is going on now as it is in characterizing what was going on in early work in experimental philosophy.

I think your own research on free will is a nice example. Your first papers on the topic serve as a real paradigm case of what people mean by the 'positive' program. In those early papers, you guys suggested that there was some reason to respect people's ordinary intuitions and that the fact that people showed compatibilist intuitions in these studies provided some reason to think that compatibilism might actually be correct.

But then your subsequent work with Feltz & Cokely adopts a more complex approach that is not quite as easily classifiable. There, you suggest that different groups of people have different intuitions (though without thereby suggesting that people's intuitions are irrelevant to the problem), and you offer the idea that people might confusing determinism with fatalism (proposing a potential confusion though again without saying that intuition per se is a flawed approach to the debate).

Of course, it might well be possible to classify your later work in some way using the original trichotomy, but I was thinking that if we just wanted to get a good understanding of what was important about this work, it wouldn't actually be very helpful to come into it with a distinction between positive and negative.

In any case, I certainly agree with the suggestion that a number of people have made that this is all just a matter of degree and that the categories might still prove helpful to a certain degree, even while proving less helpful than they did at an earlier time.

Josh May


Very interesting post. I think you're quite right that the proliferation of work in x-phi has led to more difficulty in categorizing, including using the old categories. One thing I do worry about is what seems like equating usefulness or helpfulness of a distinction with achieving a high percentage of current papers in the resulting categories.

My sense is that a categorization scheme is helpful depending on one's goals or what it does for the given issue at hand. Sometimes, for example, I still find it extremely useful and important to highlight the old positive/negative distinction when discussing x-phi with people who aren't so familiar with it and, especially, wary of it. Not too long ago I learned of a philosopher coming to not loath x-phi because he discovered it was not entirely negative (in your sense above). A large number of people in the philosophical community still suffer from this false belief.

On the other side of things, I'm not so sure what goal your standard serves. Why would it be especially helpful to find a categorization that gets a high percentage of papers included in it? I'm not exactly skeptical of there being such a worthy goal. I'm more just wondering if you could say a bit more about what you have in mind.

Josh May

Yikes, I should probably clarify what I took to be the false belief. It's the belief that most, if not all, of x-phi is negative. I take it that it's quite important to dispel such beliefs.

Joshua Knobe

Hi Josh,

You are completely right. Just as you say, it isn't really the percentages per se that are directly relevant.

I guess the real issue is that if we want to get an understanding of what people are actually doing, we might proceed by looking at all of the experiments from the past few years and seeing whether they seem to fall naturally into a number of distinct clusters. In general, it might seem odd to think that the best way of dividing these studies into clusters would be to put, say, 99% of the studies into one cluster and 1% into the other, but you are exactly right to say that it is possible that this sort of classification could turn out to be the most illuminating one.

Be that as it may, I don't think that the positive/negative distinction actually is very illuminating in making sense of people's recent work. If one looks, e.g., at your recent experiments on attributions of knowledge or at your experiments on weakness of will, it might be possible to say that these experiments fit naturally into a broader cluster that also includes papers from a number of other researchers. I could imagine various different ways in which one might do this, but in my view, it wouldn't be especially helpful to try to understand these experiments of yours by classifying them as being either positive or negative. This sort of classification doesn't seem to me to be one that actually brings out what is important about what you are doing or how your work relates to that of other researchers.

Josh May

Good points, Josh. I think you're spot on in indicating that there is so much new work that is going beyond the old major categorization scheme. And, to my mind, this is a wonderful development. While I think various ways of carving it up might be helpful for different situations, it's good to see that x-phiers are pushing the envelope in this respect. The field has simply exploded and rapidly evolved in so many welcome ways!

Tamler Sommers

I propose a new two-part classification scheme: Experimental philosophy that doesn't expose the hypocrisy and all-around iniquity of moral philosophers and experimental philosophy that does.

Antti Kauppinen

Tamler - great idea, that would split it about 50-50!

Jonathan Livengood

What do you all think about the following version (or slight expansion?) of Justin Fisher's classification scheme:

Experimental philosophy work falls into one of five categories:

1. Experimental Minimalism. These people are mostly interested in finding effects without any other philosophical axes to grind. Call them effect junkies if you like. (I think a lot of straight experimental psychology falls into this category. Depending on how widely you want to construe "experimental philosophy" -- or how sensitive you are to its history, you might include a lot of other experimental science as well.>

2. Experimental Rationalism. Assume that ordinary philosophical methodology is the method of intuition -- try to provide necessary and sufficient conditions for the application of a concept or term, support theories with intuitions about cases, attack theories with counter-examples (which are just negative intuitions about other cases). Claim that the method works fine iff there is widespread agreement about the cases. Supply evidence of agreement experimentally. This is (I think) what people usually have in mind with the "positive program."

3. Experimental Anti-Rationalism. Agree with the rationalists about the ordinary method in philosophy, but be less optimistic about its chance for success. Show by experiment that intuitions are not so widespread and that they often differ along theoretically salient dimensions (like gender, SES, culture, political affiliation, geographical region, etc.).

Positions 2 and 3 might be seen as ends on a continuum of positions that differ only in how optimistic they are. Or one might think that 2 and 3 categorize specific projects and that there is a more neutral position that accepts the assumption about how philosophy ordinarily goes and then simply asks for every philosophical domain, "How much agreement is there in this domain?"

4. Experimental Naturalism. Fill in the descriptive side of Quine's naturalized epistemology experimentally. Attempt to characterize psychological and neurological mechanisms of human thought. This includes uncovering various cognitive biases that people have.

5. Experimental Pragmatism. (I'm not real happy with this label, but whatever.) Ask why people have the concepts and play the language games that they play. What purposes / goals are served and how could they be served optimally? Assume that we can get a better handle on such questions through experimentation.

Of course, these are much too brief thumbnail sketches, but what do you all think?

Damian Szmuc


I really like your five possibilities-distinction. But I think that like in the other distinctions already made, I would prefer to combine at least two of them -in yours, the 2nd and the 5th. I feel that we could go back to the what Ben Gadd said: maybe all the inquiries divide themselves in the ones that ask for positive/negative facts about 1st order issues on philosophy (intuitions, rationality, etc.) and questions that look for positive/negative results about meta-philosophical topics (like justification of theories or something the like).

Now that I'm thinking about it, it may be possible to do an experiment to test this side effect (positive/negative) in philosophers, asking them about their feelings wheter the results of one or the other order-research are positive or negative. But that is just an idea -and I would really like to join the testing team if one of you guys decide to do this.

May this comment serve as a greeting from the University of Buenos Aires, in Argentina. We're reading very carefully your papers and books, hoping to meet you at some place of the world, some day. Hopefully soon!

The comments to this entry are closed.

3QD Prize 2012: Wesley Buckwalter