The Knobe Effect

As an avid reader of Language Log, my interest was recently piqued by a commenter asking for a linguist's eye-view on the "Knobe Effect":

"Speaking of Joshua Knobe, has any linguist looked into the Knobe Effect? The questionnaire findings are always passed off as evidence for some special philosophical character inherent in certain concepts like intentionality or happiness. I'd be interested in a linguist's take. If I had to guess, I'd say the experimenters have merely found some (elegant and) subtle polysemic distinctions that some words have. As in, 'intend' could mean different things depending on whether the questionnaire-taker believes blameworthiness or praiseworthiness to be the salient question. Or 'happy' could mean 'glad' in one context but 'wholesome' in another, etc…"

Asking for an opinion, eh?  When do I not have an opinion?  (To be fair, it happens more than you might expect).

But of course, I do have an opinion on this, and it's not quite the same as the one articulated by Edge.  This post is a long one, so let me offer a teaser by saying that the questions at stake in this are : What is experimental philosophy and is it new?  How does the language we speak both encode and subsequently shape our moral understanding?  How can manipulating someone's linguistic expectations change their reasoning?  And what can we learn about all these questions by productively plumbing the archives of everyday speech?

For those who are not familiar, Joshua Knobe is an up-and-coming 'experimental philosopher' at Yale, and is well-known for his experimental work looking at how we interpret a person's actions depending on linguistic context.  The idea underpinning his approach is that we can better understand philosophical concepts if we look at how people use and respond to them in practice.  Many of these experiments focus on intentionality : i.e., in what contexts do we say that a person acted intentionally, and in what contexts unintentionally?  Based on these findings, Josh wants to claim that he has discovered something 'deep' about the nature of theory of mind, intentional action, and moral judgment.  But has he?  I'd argue that he's discovered something about how we use certain words and what we take them to mean.  Is that deep?  Perhaps!  Read on -- and you tell me.

There is one thing I'd say first though, which is that while Josh's approach is widely taken to be innovative or revolutionary, it's almost certainly not. Wittgenstein proposed this method of investigation in the 1930's, and Chomsky roundly denied that linguistic research could tell us anything about these kinds of 'philosophical' questions in the 1960's, in response to an enthusiastic outburst by Zeno Vendler.  (See also J.L. Austin and Gilbert Rile)  So it's been 'a thing' -- or a 'thing to contest' -- for a while now.  And I'm not sure it's even proper to say that it's just now experiencing a resurgence;  any cognitive scientist working on the nature of language, language learning, or concepts, is almost certainly engaging with philosophy (and philosophical questions) in a similar, experimental way.

When philosophers use a word -- "knowledge," "being," "object," "I," "proposition," "name" -- and try to grasp the essence of the thing, one must always ask oneself : is the word ever actually used in this way in the language-game which is its original home?  What we do is to bring words back from their metaphysical to their everyday use.  --Wittgenstein, Philosophical Investigations

That quibble aside, Josh's is certainly a clever means of investigation, and one that I would endorse.  It's really the various conclusions that Josh entertains that I find a mite strange.

This is because I think there is a fairly simple (and mechanistic) explanation for the "Knobe Effect."  The effects he gets in his experiments come straight out of corpus data (i.e., reams of data about how we use words in everyday speech).  That may sound like so-much Greek to you, so let me first begin by giving you some examples of the kinds of experiments Josh conducts.

In one of his best known papers, he asked bystanders in a public park to read one of the following two stories.

"The vice-president of a company went to the chairman of the board and said, ‘We are thinking of starting a new program. It will help us increase profits, but it will also harm the environment.’  The chairman of the board answered, ‘I don’t care at all about harming the environment. I just want to make as much profit as I can. Let’s start the new program.’  They started the new program. Sure enough, the environment was harmed."

"The vice-president of a company went to the chairman of the board and said, ‘We are thinking of starting a new program. It will help us increase profits, and it will also help the environment.’  The chairman of the board answered, ‘I don’t care at all about helping the environment. I just want to make as much profit as I can. Let’s start the new program.’  They started the new program. Sure enough, the environment was helped."

Josh then asked the bystanders to rate whether the chairman had intentionally harmed / helped the environment on a scale of 0-6.  The results were intriguing:  Those who read the 'harm' story were eager to blame the chairman for the ill-effects wrought on the environment (and tended to agree that he had 'intentionally' harmed the environment).  Those who read the 'help' story, on the other hand, weren't eager to award the chairman credit (and tended to agree that he hadn't 'intentionally' helped the environment).  In other words, even though the stories were virtually identical, intentionality was far more likely to be attributed to the chairman in the 'harm' situation than in the 'help' situation.  (The differences were highly significant; p <.001)

Josh has since replicated this effect in several similar studies.  The question is -- what does this tell us?  In the original experiment, Josh concluded mildly that

"there seems to be an asymmetry whereby people are considerably more willing to blame the agent for bad side-effects than to praise the agent for good side-effects. And this asymmetry in people’s assignment of praise and blame may be at the root of the corresponding asymmetry in people’s application of the concept intentional: namely, that they seem considerably more willing to say that a side-effect was brought about intentionally when they regard that side-effect as bad than when they regard it as good."

Not much I disagree with there; but we'll get to his more ambitious conclusions in a moment.  First, though, let's take a detour --

A Complimentary Vein of Research

At the time that I became familiar with Knobe's research, I was hard at work on an article with Prof Plum and Prof Teenie Matlock about lexical priming effects on temporal reasoning.  --Which is a fancy way of saying that we were looking at how we could mess with people's understanding of time by having them read different motion verbs in context.

The experimental design was similar in style to Knobe's.  In four experiments, we had people read a series of virtually identical 'priming' sentences and then had them answer the following question : "Next Wednesday's meeting has been moved ahead two days.  Which day is it on now?"  (Stop.  Pause for a moment.  Which day is it?)

Trick question.  This question is ambiguous because the answer could either be Monday or Friday, depending on how you're conceptualizing time.

Don't believe me?  Try this on for size:

First, imagine that time is like a conveyor belt and that you are standing in place, as it slips past you.  If you conceive of time this way, then the future is a point up ahead, moving ever closer to you.  Adopt this perspective, and you'll almost surely answer Monday because moving time 'ahead' will move it closer towards where you are now.

Now imagine that the conveyor belt stops, and you are the one moving through time.  In this case, moving 'ahead' in time will move you farther along into the future.  If you imagine it this way, then you should answer Friday.

These are known as "time-moving" and "ego-moving" metaphors, respectively.  See now?

(Don't worry if you don't right away.  It's kind of like those ambiguous duck / rabbit pictures; it can be hard to see one once you've spotted the other.)

But back to the experiment -- we know that when faced with that ambiguous Wednesday question, participants split pretty evenly between Monday and Friday.  What we wanted to see was whether we could push our participants toward one answer or the other, simply by having them read a short series of sentences first.  As in the Knobe experiments, the sentences were virtually identical but differed on one critical word.

In a particularly illustrative test of this, we had participants read sentences that differed on the motion verb -- for some it was "comes," for others "goes" (e.g., "The road comes all the way from New York" versus "The road goes all the way to New York").  They then answered the ambiguous time question.  What we found was that participants who read "comes" skewed heavily toward Monday responses, whereas participants who read "goes" skewed heavily toward Friday.  --Which was exactly what we expected.

The question is : why?  Or perhaps -- how?  How were we able to manipulate their temporal reasoning in this way?

Simple!  By manipulating their expectations.

In English, when we use the word "comes," we typically follow it up with 'past-looking' words, like sooner or before (e.g., "I hope summer comes early this year" or "My station comes before hers"). But with "goes" we tend toward just the opposite pattern; we favor 'future-looking' words like after or later ("he goes later in the heat" or "her talk goes after his").  If we look at the words that follow "comes" and "goes" in a large record of human writing and speech -- say, the Contemporary Corpus of American English (COCA) -- then we find that these differences aren't trivial in English.  While the log ratio of future to past-looking words for "goes" is 2:1, for "comes" the trend reverses, with a log ratio of 1:2.  In plain English, this means that "goes" is future-biased, while "comes" is past-biased.

Still, you might be left wondering  -- why should our participants be sensitive to this information?

Over the past couple of years, there have been a number of truly groundbreaking studies showing that we probabilistically track how words are used in speech (we know whether this word is used with these words or with those, and to what extent these and not others).  These 'usage patterns' inform our expectations about what's coming next in reading or conversation.  So, for example, if I strike up a new topic by asking, "What are the effects of doing ___?" you've already begun probabilistically anticipating words like "drugs," "business" and "exercise," (and I can bet you're not thinking words like "laundry," "research" or "kung fu").  Or, to give another example, should I say, "I'm coming down with a ___," then you'll be expecting words like "flu," "cold," "virus, "and so on.

We know that people are able to track this kind of information, because if we throw in a curveball and say "What are the effects of doing Youtube-style videos?" or "I'm coming down with a bottle of beer later," then we can watch their brains registering a pretty mean N400 (surprise!) spike, as they work a bit harder to compute what was just said.  (Impressively, the size of the neural spike correlates well with people's reported expectations about what should fill that slot.)

Flipping back to the 'time' research for a minute, it's pretty clear that we should be sensitive to the words that tend to follow a given word, and even to whether that word tends to hang out with future-looking words or past-looking words (i.e., whether it's future or past biased)  From this vantage point, what the "comes / goes" experiment is telling us, is that we can cleverly push our participants towards "Monday" or "Friday" responses simply by manipulating their linguistic expectations.  Get them thinking "past" they'll answer Monday; get them thinking "future," they'll answer Friday.  Now, if you're not impressed by this example --which may seem rather intuitive, on inspection-- you may be more impressed by the next experiment we did.

In that experiment, we had participants read a single sentence ("X trees run along the edge of a driveway") before answering the ambiguous Wednesday question.  The only word that differed between participants was 'X', which was always a number word (either four, eight, ten, eleven, twelve, nineteen, twenty, over eighty, 0r a hundred).  Because number words are frequently used with time words, and because time words are future-biased, we expected that all of these words would bias participants toward a Friday response.  The question is whether we could account for the strength of that bias.

So Prof Plum built some handy models of what linguistic expectation might look like using COCA, and then tested the fit with our priming data.

To examine whether lexical prediction might provide a [good account for our data], we compared the responses provided by our participants given the various number primes to our models of the linguistic expectations the numbers could be expected to produce. As we expected, there was a good fit between the predicted priming and the degree of bias exhibited in the empirical data. The predictions of Model 1, in which we sought to account for the way that different time words might be expected to produce different degrees of future bias, correlated well with the pattern of data produced by our participants (r=.76, t=3.09, p<0.01). Perhaps surprisingly, however, the simpler model (2) performed even better (r=.92, t=5.46, p<0.001). Indeed... if only the proportion of time words amongst the 10 most frequent words in the distribution following each number word is considered, this correlation increases (r=.96, t=8.05, p<0.0001).

(Won't lie -- I do find that r=.96 to be mildly satisfactory).

The bottom line is : We have implicit knowledge of how words are used in speech that comes from a lifetime of exposure to the distributional patterns of words in our language.  In reading or listening, this knowledge shapes our expectations and understanding of what we're taking in.  As scientists, we can subtly, or even powerfully, manipulate people's expectations and understanding of a given question or task by using particular words in particular ways.  And even better, we can predict -- with no little accuracy!-- the effect this is going to have on our subjects.

(Didn't advertisers figure this out a long time ago?)

Returning, then, to Knobe Effects

Having just finished the write-up of these findings when I read the "harm / help" paper, I was immediately struck by the thought -- what if Knobe effects can be explained in a similar way?

Curious, Prof Plum and I set out to investigate.  To do so, we started out with a simple question -- how do we talk about 'help' or 'harm' in terms of intent? To get a measure of this, we looked at the words preceding 'help' and 'harm' in COCA, and calculated how many of these words expressed intent.  (Intentional words include 'will,' 'can,' 'may,' 'would,' 'could,' 'should' and so on).  We wanted to know how often we say things like "I could help her" versus "He will harm the project."

The difference was striking (and optimism-inspiring!):

It turns out that we talk far more about intentionally helping than intentionally harming or hurting (another word we tested).

To be specific : if we account for frequency differences between 'help' and 'harm,' the proportional difference is about three-fold; if we simply look at raw frequency counts, that number skyrockets to a fifty-odd difference. Taking either figure into account, the fact is that we have considerably more practice talking about "help" in an explicitly intentional way than we do "harm."

At first glance, this might seem like a surprising result.  If we often talk about 'help' in an intentional way, then why didn't we interpret the chairman's actions as intentionally helpful in Knobe's scenario?  Well -- quite simply -- it's because we have more practice with it.  We know that 'help' is often used intentionally in certain contexts, which clearly don't match the scenario Knobe spells out.  With "hurt," on the other hand, we have far less practice with talking about it in explicitly intentional contexts.  This means that we should be less discriminating in how we apply it.

But of course, there's more to it than that.  It's also the case that we're far more likely to mention that something is unintentional when it's produced ill-effect.

Here are the results of a simple Google search comparing usage of "didn't mean to..." with "meant to..."

We can see from usage of "didn't mean to..." that when we're emphasizing that something is unintentional, we're far more likely to be talking about hurt / harm than help / support.  (Just to make the contrast even more obvious : the phrase "didn't mean to hurt" captures fully 5% of all usages of the word 'hurt,' whereas "didn't mean to help" accounts for just .001% of usages of 'help.')  Looking back to the chart -- notice how this trend does an about face when we look at usage of "meant to..."!

Even this chart doesn't fully capture the differences though.  As I was double-checking the counts, the song "Do you really want to hurt me?" came to mind.  Hmm.  Do I?  What would happen to the frequency counts if I took out all instances of "never meant to," "not meant to," "wasn't meant to," and "weren't meant to"?

Wow.  So once we clean that up, the "meant to hurt" counts drop by almost 80%, while "meant to help" count drops by, oh, a respectable 4%.  This data is screaming : intentionality is the bastion of aid, not injury.  Or, to put it somewhat differently : we like to take credit when a good act is intentional, and, equally, we like to avoid blame when a bad act is unintentional. In either case, we're apt to point it out!

All of which brings us back to Knobe's experiments.  There is one, absolutely vital thing to note there : neither intentionality nor unintentionality is made explicit in the scenarios he presents.

What's so delightful about this, is how our expectations of what should be made explicit drive us in opposite directions in the case of 'help' versus 'harm.'  So, with regards the "chairman-environment" scenario, here's the rub:

--Since the good act hasn't been explicitly noted as intentional, we're unlikely to see it as intentional, since we expect to see that attribution when it's relevant.

--At the same time, since the bad act hasn't been obviously noted as unintentional, we're unlikely to see it as unintentional, since, once again, we would expect to see that clarified if it were!

This provides one alternative way of accounting for Knobe's data.  Of course, you might think (rightly) that this is simply a re-description of his findings in terms of word meaning and use.  For instance, he says :

By systematically varying aspects of the vignettes, researchers can determine which factors are influencing people’s intuitions. It can thereby be shown that these intuitions show a systematic sensitivity to moral considerations.

In fairness, there is nothing I disagree with here -- 'Help' and 'harm' are, to my understanding, 'moral' words.  So if their usage reflects moral considerations, my analysis (and Knobe's conclusions) are entirely consistent.

However, I would press that by offering a more mechanistic account of what is happening in this experiment, we can clear away some of the confusion about what is motivating these results.  For example, Knobe lays out these possible explanations for his data:

One view holds that the emotional reactions triggered by morally bad behaviors can distort people’s theory-of-mind judgments.  A second view holds that moral considerations play a role in the pragmatics of  people’s use of certain terms but not in the semantics of their theory-of-mind concepts. Finally, a third view holds that moral considerations truly do play a role in the fundamental competence underlying people’s theory-of-mind capacities.

To this, he adds a fourth possibility:

These results provide at least tentative support for the view that the effects can emerge even in the absence of emotional responses, and some researchers have recently suggested that the effects might be due, not to an emotional bias, but rather to an innate, domain-specific ‘moral grammar.’

Of course, this is Hauser's speculation, not Knobe's.  But -- an innate, domain-specific moral grammar?  I certainly wouldn't draw that conclusion from all this.

More seriously, the analysis I've offered suggests something very simple : that the way we understand and attribute intentionality has to do with our particular linguistic expectations about how "good" or "bad" actions will be described within a given narrative frame.  Far from being 'innate' or 'universal,' these expectations may be (and are even likely to be) culturally contingent.  Caitlin Fausey, a graduate student wunderkind, formerly of Lera Boroditsky's lab, has done some fascinating work on how causal attribution and blame differs between languages, such as English, Spanish and Japanese.  Her work shows, quite clearly, that the way we use language to describe events can significantly impact our reasoning about (and memory for) them.

From the perspective of -- well, this armchair at least -- Knobe's work can be seen as adding to that literature; a curious and important contribution, no doubt.  His results also tell us something about how the language we speak both encodes, and subsequently shapes, our culture's moral perspectives and expectations.

Is this deep?

I'd say it's unexpectedly so.


Knobe, J. (2003). Intentional action and side effects in ordinary language.  Analysis, 63 (279), 190-194 DOI: 10.1111/1467-8284.00419

Knobe, J. (2005). Theory of mind and moral cognition: exploring the connections. Trends in Cognitive Sciences, 9 (8), 357-359 DOI: 10.1016/j.tics.2005.06.011

Ramscar, M., Matlock, T., & Dye, M. (2010). Running down the clock: the role of expectation in our understanding of time and motion. Language and Cognitive Processes, 25 (5), 589-615

See also

1.  The Chomsky-Foucault Debate (Hat tip to Ben H.)

2.  The following papers on the "time" experiments : the pioneers, McGlone & Harding (1998); see also Boroditsky (2000); Boroditsky & Ramscar (2002); Matlock, Ramscar, & Boroditsky (2005).  There are many more besides; Ramscar, Matlock & Dye (2010) provides a brief review.

24 responses so far

  • Terry says:

    Not to be disrespectful to Knobe, but this looks an awful lot like armchair philosophy. The statements he tested were neither culture-, morally- or intentionally neutral.

    The two cases are clearly not ethically identical. In the first case the executive is clearly being alerted to a condition that would indicate caution or at least inaction. In the help case, there is no indication. Further, the executive clearly states his intention - to make profits. The assumption of guilt by participants is centered on the idea that he/she took positive action on his only declared intention when inaction or counter action would be called for to locate a morally positive result. In the help case, the results were not dependent on his/her intention. It was simply a by-product.

    There are other issues here as well - an executive of a corporation conjures a whole different set of expectations than a doctor or a schoolteacher. As you point out, words do not exist in isolation, some of the probabilities we select as we prune down the tree of an advancing grammar is based on culture as well and our expectations for functioning in the society that defines the culture. Didn't Wittgenstein say as much when he said we would not understand a lion if he were to open his mouth and speak to us?

    I'll have to look at the source material on this to get a better idea of what Knobe is up to, but from this superficial reading, it doesn't seem deep so much as it seems poorly designed.

    • melodye says:

      I think there's a lot of interest here, but to be fair, the analysis I offer (above) is probably not how Josh would interpret or explain these results at all. I'm rooting about for a depth in the work that I find to be authentic and empirical (and not, as you say, armchair philosophy).

      Isn't Wittgenstein lovely? I think my favorite philosophers are always the ones that manage to read like poets.

      • Terry says:

        Wittgenstein is wonderful, but dense and occasionally opaque. For 25 years I have been coming back to him and still am not quite sure I..get it 🙂 For just the joy of reading, I love Borges. Roland Barthes, although I thought his goals were futile, can be pretty beautiful on the printed page as well.

        I'm eager to read Knobe. I hope he surprises me.

  • Dean Eckles says:

    While I found your approach interesting, I'm surprised you did not directly discuss any of the many subsequent studies in this paradigm that, I think, have made it very difficult to maintain Knobe's original explanation. There is a "Knobe effect" all right, but I am not sure it has anything to do with moral criteria for an event being intentional (or intended, etc.).

    I haven't reviewed this literature recently, but the explanation that appealed to me most (and wasn't falsified by what I've read, I think) is that these vignettes result in people attributing different levels of consideration of (e.g.) the effects on the environment in the CEO's decision-making. That is, we could say that what drives the intentionality judgments is whether the effects on the environment entered into the (imagined) pro/con calculus of the agent. Furthermore, in the "harms" case, a different weight on the effects of the environment in his calculus could conceivably change the decision: the decision was made is a way that is responsive to caring about the environment. But in the "helps" case, the decision is completely unaffected by how much the agent weights effects on the environment in decision-making.

    (See the papers Knobe links to here:

    • melodye says:

      Which of the two dozen papers should I read?

      Again, what I think is interesting here is the drive to explain this in terms of people's "conscious" decision making processes. As an alternative, there's a very simple way to explain the results, which is to say -- OK, people have these linguistic expectations about how "help" and "harm" are talked about: i.e., people assume that

      1) help that is intentional will be described in intentional terms &
      2) harm that is unintentional will be similarly specified.

      So, when you give them a scenario in which the presentation defies these expectations (help is not specified as intentional; harm is not specified as unintentional), they conclude the opposite.

      This comes directly out of how we use these words in everyday speech, and has little to do with 'moral criteria.' It's simply the way that we use supposedly "moral" words, like harm / hurt as opposed to help / support.

      To be clear, I'm agnostic about debates within philosophy about "morality" and "intentionality" and so on. I'm just adopting these words as a means to investigate how we respond to language. Philosophers (like Josh) seem to want to make important claims about morality and intentionality based on experiments like this; my interest is in how we can manipulate how people reason and respond within specific frames. It's clear from the time experiment that we can manipulate people's temporal reasoning; why not also manipulate their moral judgment?

      This doesn't tell us anything particularly profound about morality as such. It does tell us something quite fascinating about language, however.

      • Dean Eckles says:

        I'm glad you put "conscious" in quotes, since I would not want my comment to commit me to anything about the process being conscious.

        What follows is a basically just a guess -- or maybe a hypothesis. These effects are much too large to be primarily explained in the way you suggest, at least as I understand it. There are many pairs of words that exhibit the same magnitude of covariation with intentionality judgments, but do not produce these effects when swapped in or out as in this experimental paradigm.

        (It may be a fun exercise to read poverty of the stimulus into the above argument.)

        I would recommend Malle's work on this. See

        My considered view on these cases is something like: as usual, all the causes are in play.

        • melodye says:

          A couple of things : First, I think poverty of the stimulus arguments are, in general, bogus, and rely on some pretty questionable reasoning. (You can read Fiona Cowie's encyclopedia article on this, "Innateness in Language," and also many of Geoff Pullum's articles; we've also done empirical work on this question, showing that supposed puzzles are explicable in learning theoretic terms).

          Second, you say "There are many pairs of words that exhibit the same magnitude of covariation with intentionality judgments, but do not produce these effects when swapped in or out as in this experimental paradigm." But that's an empirical question, not a statement of fact, and as far as I know, no one else has done this kind of corpus work. How then would they have shown this to be true? Our intuitions are not always a good guide to how language actually works.

          Finally, you write that your considered view is that "all the causes are in play." All of what causes? If you can account for the data in simple linguistic terms, what other causes do you need to posit?

          I find this sort of reasoning hard to stomach, given that I face it in review all the time; "so, your learning model accounts for all the data here, but what about all the other causes? like a universal grammar? It simply *must* be more complex!" But wait -- why? Because linguistics and philosophy have wanted to make it so?

          Turning to Malle, then : he says, "[The] influence of intentionality on blame involves a complex conceptual framework, as intentionality judgments rely on multiple necessary conditions" --Or maybe, instead, intentionality judgements rely on our expectations about how words like 'intention' and 'blame' are used in language. There's quite a lot of research in psychology, by the way, suggesting that humans don't "logically" reason through things (in the manner of a formal logic) unless explicitly trained to do so. What philosophers want to posit and what people actually do are often way out of sync.

          It's worth noting that my description above is not qualitatively different from Malle's; it differs in that it is explicitly mechanistic. For example, Malle says "People generally expect others to prevent negative outcomes and to promote positive ones" and notes that, " The agent’s professed indifference was interpreted as moderate desire for the outcome in the negative case but as virtually no desire in the positive case." Compare to what I said above : "[In language], we like to take credit when a good act is intentional, and, equally, we like to avoid blame when a bad act is unintentional." They're not identical statements, but the principle motivating them is similar (see Malle's Exp 1, 2, 3). Note that in the follow-up experiments, the language changes (e.g., in Exp 4, the study becomes : "It would be unfortunate if the environment got harmed. But my primary concern is to increase profits. Let’s start the new program.")

          I'm not sure if you meant Malle's work as a foil; he seems to be saying something highly similar, but in philosophy-speak. My preference is to keep things as simple and well-defined as possible, but otherwise, I'd have to say we agree.

          • Dean Eckles says:

            The reference to poverty of the stimulus was meant as a joke, given the previously posted comic.

            Lack of empirical evidence is why I stated it as a hypothesis. The same story cuts the other way, of course. But I do think its reasonable to expect that there are other words with similar empirical associations but that don't have effects this large. I would think some formal model comparisons would be in order if there are dueling theories. To "explain" all the variation, I'm just saying that we would need to include many of these causes in our model.

            By the way, though Malle has waded into experimental philosophy, he did his PhD in psych here at Stanford. Much of his research has aimed to substantially revise or replace traditional attribution theory in social psych.

            I suggested this other work because I don't think that Knobe's "deep" story has found much empirical support in the growing literature on this. Since his first papers, replies have argued for more surface explanations, including some based on pragmatics stories. So, "yes, you are not alone" in not buying his theory was meant to be part of the message.

  • Seth says:


    I think I undertand the argument Melody. Looking at the following qoute however:

    '1) help that is intentional will be described in intentional terms &
    2) harm that is unintentional will be similarly specified.'

    I think begs a question.

    Why does everyday language tend specify harm that is unintentional in intentional terms?

    When we use language to describe harm that has occurred from our own personal actions are we not less likely to use intentional terms?

    Just because general language usage is 'so' in the qouted example doesn't mean language itself is root cause of the dynamic at play, does it?.

    Couldn't the common language use be the outcome of the way we attrbiute intent to others as opposed to ourselves?

    As may be apparent from my quetions I am not especially knowledgable in linguistics.

    • melodye says:

      Great question. I think I'm struggling to articulate this bit, because I don't want to overstep what I think we can say about findings like these.

      But to be frank, this is where I think it becomes really interesting right -- the language we speak enshrines - in a very important sense - something about our conventions (or 'norms'), both for 'moral' and 'intentional' reasoning. (It seems to be both a reflection of those shared norms, as you say, and, in a gentle sense, an implicit constraint on thinking about them). So, you ask :

      Couldn’t the common language use be the outcome of the way we attrbiute intent to others as opposed to ourselves?

      This is the million dollar (chicken or egg) question. It could be that there's a 'universal moral grammar' - as Hauser proposed - or maybe something quite a lot less exciting than that; maybe simply a universal characteristic of how humans, across language, tend to attribute intent or understand causality. (See my discussion of why seemingly universal characteristics don't always necessitate innate explanations).

      The reason I think it's not likely to be that is because of cross-linguistic work showing that there are differences in how people understand causation -- and subsequently how they attribute blame -- depending on the language they speak, notably by Caitlin Fausey with Lera Boroditsky, and by Dan Slobin. What's more, a lot of the work I do with Prof Plum suggests that many of the causal relationships we care about (and attend to) in the world have to be learned. Different languages can have very different ways of telling us which relations to attend to, and which parts to focus on.

      • Seth says:

        Thanks very much for the reply.

        I am not arguing for universal moral grammar or innate morality in general.

        Lets say hypothetically that on a particular population Knobes rule had no effect on judgement of intent, and that this was consistent with that populations language usage. We still wouldn't know the causal factor behind the intentional judgements. Could possibly be purely the language I guess, but I would be interrested in what evolutionary and or cultural factors might be behind the language usage.

        As a quantiative researcher I am interested in bias in general and trying limit the ways in which it can trip us up. Identifying the root cause is always difficult. I would bet that language development and moral attitudes likely fed off each other in a loop like relationship likely in response to other factors (just vague baseless speculation). I think what is most important is simply becomming aware of where bias is likely to exist and of the ways that it is likely to influence our beliefs and behaviors.

        I'll check out the links,

        thanks again

  • GuessHandsOn says:

    Very interesting, M-dawg...

    Nonetheless, maybe there is a way to interpret your data without 'explaining away' Knobe effect. For instance, you say that:

    "Since the good act hasn’t been explicitly noted as intentional, we’re unlikely to see it as intentional, since we expect to see that attribution when it’s relevant.

    At the same time, since the bad act hasn’t been obviously noted as unintentional, we’re unlikely to see it as unintentional, since, once again, we would expect to see that clarified if it were!

    Why do we expect good acts to be explicitly noted as intentional and the opposite for bad acts? Maybe there is a morally normative component in intentionality attributions; that's why.

    Your message seems to be that: usage of words like harm, help, hurt => Knobe effect. (=> is the arrow of causality or explanation.)

    I'm saying, why can't it be: Knobe effect => usage of words like harm, help, hurt.

    One thing you and Ramscar can do is, perhaps, test the Knobe effect on children. Since children have not been exposed to a lot of language, if they do exhibit the Knobe effect, then "Knobe effect => usage of words like harm, help, hurt." There is a moral component in intention attributions.

    Of course, another datum would be cross-cultural studies. (But I think they have been done and a lot of cultures exhibit the same effect.)

    Hmm... I wonder what Chomsky would think. Perhaps, I should email and ask...

    • melodye says:

      I mean -- you're right that I'm probably being a bit bold to point the causal arrow that way. I think it probably goes both ways (see response to comment above).

      I would be curious to see any comparative studies on this. As I mentioned, there is a lot of cross-linguistic work in a similar vein showing substantial differences. One way to test this hypothesis, would simply be to look at a variety of corpora across different languages and examine whether all the languages tested show a similar distribution (i.e., manner of talking) for words like help / harm. If you could find one that didn't, then you could test whether the "Knobe Effect" really holds up - i.e., is a universal characteristic about how we morally reason, somehow independent of language.

      My hunch is that it's not.

      W/r your suggestion : children actually prove remarkably sensitive to how words distribute in their languages, so it doesn't seem like testing them would be very helpful. (See e.g., Bannard & Matthews, 2008 and our replication of it).

  • Firionel says:

    I was already thinking this when I read your post on concepts, but now I feel it is starting to get urgent: So when will you be writing about social constructivism?

    The problem here quite obviously runs veeeerry deep. Language (insofar as language is the origin of the effect described) apparently has normative aspects in its descriptions of intent and causality, and my gut feeling is that that phenomenon is way more widespread than one would expect at first sight. (I think I read an article about a related problem in the description of causation by omission recently, but I can't quite remember for the moment.)

    And of course Seth seems to be right when he says that just discovering that language seems to be what's guiding our thoughts here is somehow begging the question: language itself is an expression of human thought processes, so 'language is like that' would not seem to be a satisfactory explanation. (Unless on believes in innate language, but I think we are all pretty certain you are in no two minds on that question....)

    • melodye says:

      With regards to social constructivism -- what aspect do you want me to take on? I have written (briefly) in this regard on gender and am planning to do a follow-up. I'm not sure if you want me to discuss the theory, generally; how our work informs that theory; or on something more specific (like gender, desire, and so on).

      You also might be interested to read the conclusion to "Running down the clock." Here it is, in excerpted form :

      There are many reasons to believe that human knowledge cannot be easily or neatly bounded into “domains” that words can simply refer to (Wittgenstein, 1953; Lakoff & Johnson, 1999; Murphy, 2002). This in turn presents a massive obstacle to simple, unitary characterizations of the nature of human understanding. While at first blush it might seem that more abstract domains simply borrow conceptual structure from more ‘literal’ experiential domains, the reality is almost certainly more complex. It is likely that even the structure of experiential domains derives in part from the structure of language. As Wittgenstein (1953) famously pointed out, seemingly straightforward experiential concepts, such as the kinds of things we call “games,” do not exist independently of language and culture. Rather, it appears that the structure of “game” comes, at least in part, from the way that the word game is used (Wittgenstein, 1953).

      Our proposal, that patterns of usage are part of the structure of “concepts,” can be seen as fleshing out at least one mechanism through which use may influence meaning (Wittgenstein, 1953). From this point of view, literal motion is not a “concept” from which metaphorical motion borrows structure, and the common language used to describe literal and metaphorical motion does not merely reflect those borrowings. Rather, peoples’ understandings of literal and metaphorical motion are the product of shared cognitive processes that generate an “understanding” when making a plan, performing an action, or answering an ambiguous question in an experiment (for example). The shared patterns of co-occurrence in talk about literal and metaphorical motion are not mere artifacts of some underlying conceptual structure, but rather make up a critical part of that structure: one that may influence the way that someone makes a plan, performs an action, or answers an ambiguous question in an experiment.

      If we accept that even our understanding of experiential concepts has a linguistic component, and that the particular ways in which people use language may impact their conceptual understanding, it follows that both literal and abstract ideas will be shaped to some degree by the way that languages are used. From this perspective, language can be seen as more than a referential code; rather, language offers a medium for structuring, encoding and transmitting cultural knowledge (Tomasello, 1999).

      This view of conceptual structure is consistent with the Wittgensteinian conception of knowledge and language “as an ancient city: a maze of little streets and squares, of old and new houses, and of houses with additions from various periods; and this surrounded by a multitude of new boroughs with straight regular streets and uniform houses” (Wittgenstein, 1953). Wittgenstein’s broader suggestion is that language itself might not be so neatly bounded; that there is no simple process that corresponds to “language,” but rather that communication is a “form of life,” involving a variety of related practices. While precisely what this larger suggestion entails is something that remains to be determined, our findings here suggest that this more radical view of the nature of human communication—and in particular of the relationship between “language” and “other” aspects of culture and cognition—may be worthy of further consideration.

      • Firionel says:

        It was precisely because you had written on gender typifications before that I was starting to feel that social constructivism would be a natural framwork in which to organize a number of the convictions you have given voice to in this space. (Goodness, what a monstrosity of a sentence. But I'm not going to redo it, so live with it.)

        I was specifically thinking about the Berger/Luckmann account of vocal sounds becoming signs of objectifivations and then typifications and the resulting interaction between language and 'common knowledge', i.e. social reality. (Shame on me, I haven't read a single page of Wittgenstein in my life...) They also talk about intentions being reconstructed from objectifications, which seemed particularly relevant in this context.

        I was just wondering whether that conceptual framework is one you feel comfortable being (at least loosely) associated with, since so far you have largely written about frameworks you reject. I guess your comment basically answered that.

  • [...] a completely fascinating discussion of language, context and its use in experimental philosophy over at Child’s [...]

  • Phil says:


    Your work with Blum & Matlock on lexically priming temporal judgments is very interesting. Here’s a variant of your experiment that you’ve inspired me to play with. “Next Wednesday’s meeting has been rescheduled. It is now moved ahead two days.” (Which day is it on?) The idiom “ahead of schedule” is very common, and I predict that priming with “schedule” will skew responses strongly in favor of Monday. What do you think?

    In your original design did you give participants the option to say “moved ahead” is just not clear? I suspect careful people (% unknown) in ordinary conversation would say something like “Did you mean the meeting was moved up two days?” Interesting that “moved up”, strongly contrasting with “put off”, isn’t unclear, or do you think some people still might be confused with “moved up” depending on the priming?

    • melodye says:


      I'd be curious about your results too! Our experiment was done as a written test, so we'd have the students read a number of items at the top of the page, and then answer the question at the bottom. (There was no option to 'correct us' or ask us to clarify, which might have changed the results). Hypothetically, you could try the same thing out on your friends and coworkers with your proposed version! I think our n was 50+ for both experiments, but the effect sizes were big enough that you could probably get something like a result from as little as 20 subjects. Please report back if you try it 🙂

      Also Prof Plum is really Prof Ramscar -- I just call him 'Plum' as a joke (it's the name of the Prof in Clue!)

  • Firionel says:

    Having re-read the post I think I can finally articulate what I wanted to ask all along:

    I seem to remember that there are languages which describe spatial location never in relative terms but always absolute terms. (In a vector space instead of an affine space, for the mathematically inclined.) The fact that we are in principle able to do something like that (say, in physics, where 'left' is almost never a meaningful concept) gives us a leg to stand upon when talking about ego-movement vs. time-movement.
    When looking at morality, the situation is different. We do not possess an ethically neutral description of the world (nor, for the most part, do we believe there is one). This puts us in a somewhat more difficult situation when talking about the phenomenon Knobe describes. Consequently I feel somewhat queasy when language is produced as an explanation. Morality and description of morality are inextricably linked, whether we like it or not, but finding a 'more important causality' seems to be a fool's errand.

    That said I was originally going to write that the Knobe effect was obviously 'real' in some sense, but I've since changed my mind. It may very well be an artifact of the description and set-up of the situation (more so than of the wording). Economists have done some research there, and it seems to be rather easy to make the Knobe effect disappear or turn around by either changing the description like so or making the people actors themselves like so.

    • melodye says:

      I would love to read the first paper, but received the following message : "Error 404 - Seite nicht gefunden." The second is very cool; thanks for sharing. I'm not sure it means that the wording has little to do with it though; all the experiments are in language, and a participant's expectations about him/herself ("you") as opposed to an unknown actor ("John") are going to be different both in language and without. (For example, my understanding of myself, as constructed in language, will have a lot to do with my usage of the word "I" -- the singular 'corpus' of my self, so to speak).

      I didn't quite understand the contrast you were trying to draw between spatial terms and morality. I was wondering if you could cash that out for me. You said : "When looking at morality, the situation is different. We do not possess an ethically neutral description of the world (nor, for the most part, do we believe there is one)." Of course, we don't ; but no one has a spatially 'neutral' description of the world either, so I'm not sure I understand the point you're making. Please explain further, because it seemed quite interesting to me (I just couldn't follow; 'language as prediction' fail 😉 ) Also, I assume you're talking about the Boroditsky / Gaby work (and others like it) re the spatial findings. But there are some reasons to doubt the strong conclusions of that research (though the experimental work is no doubt impeccable).

      Finally, *shame on me* : I haven't read the "Berger/Luckmann account of vocal sounds becoming signs of objectifivations and then typifications and the resulting interaction between language and ‘common knowledge’, i.e. social reality." But I will try to find it and read it (and if you want to be helpful in this regard, and send me papers or books -- please, please do; I am starved of them at the moment, lacking institutional journal access).

      • Firionel says:

        You are right, the link is indeed broken. I didn't notice that, since I had a copy of that article on the hard drive and had just used GoogleScholar to come up with a link. Here however is one that works, and it's ungated to boot. You really might want to read that one, their takeaway is this: "We find the Knobe effect without the use of the word intentionally. Therefore we conclude that the Knobe effect does not depend on language but on the economic determinants of the situation [...]." (There's room to disagree, though.)

        As for the contrast between spatial and moral descriptions, I had been thinking along those lines: I can describe 'space' as if I did not exist, that is without any reference to either myself or my own experience, whereas I cannot describe 'intention' without using my own experience as a guide. Having given it a little more thought I am willing to concede that the first part of that sentence is of dubious truth, yet I still think it might be true. (Poincaré, who I is some kind of an authority in that regard, might agree, as might formalist-realists. Fictionalists might disagree. Most linguists - I am guessing - will disagree.)
        Anyway, the difference is not as clear cut as I had initially thought.

        The book I was talking about was this one, it's a classic, but I'm not sure it's still entirely up to date. Be aware that I have no formal education in the subject, and most likely badly misrepresented it.

  • [...] 7. Explaining the Knobe Effect [...]

  • […] The Knobe Effect Scientopia: The Knobe Effect […]