Behavioristic communication

« previous post | next post »

Last week ("Joos jokes", 8/14/2018) I linked to the "Proceedings of the Speech Communication Conference at M.I.T.", published in 1950 in the Journal of the Acoustical Society of America,  and I promised to revisit "this window into a bygone age" in a later post. So today I present to you the following passage from "Introduction: A Definition of Communication" by S. S. Stevens:

Although no phenomenon is more familiar to us than communication, the fact of the matter is that this magic word means many things to many people. A definition broad enough to encompass all these meanings may risk finding itself dissipated in generalities, but for the purposes of this conference a broad operational definition of communication is, I believe, both appropriate and possible. I should like, therefore, to venture the following: Communication is the discriminatory response  of an organism to a stimulus.

He goes on to confirm that this perspective is indeed as weird as it seems:

This definition says that communication occurs when some environmental disturbance {the stimulus) impinges on an organism — and the organism does something about it (makes a discriminatory response). If the stimulus is ignored by the organism, there has been no communication. The test is differential reaction of some sort. The message that gets no response is not a communication.

This definition is broad, operational, and behavioristic. It includes the anxious clucking of the mother hen, which brings the chicks scurrying to shelter. At a different extreme it includes the modem treatise on information theory, which some people seem to read and respond to with a glow of understanding. By appealing to behavioral operations as the test for the presence or absence of communication, we explicitly forsake all concern with abstracted meanings, significations, and the like, unless, of course, these words are in turn defined in terms of discriminatory responses. In short, we stick to observable phenomena.

Are we really supposed to believe that some people's "glow of understanding" on reading about information theory is an "observable phenomenon", sufficient to characterize what has been communicated? This strikes me as a reductio ad absurdum of Stevens' behavioristic prejudices — as effective an argument as Noam Chomsky's 1959 Review of B.F. Skinner's Verbal Behavior.

Stevens then argues that this absurd idea is motivated by the desire to apply Shannon's recently-published information theory to human communication:

One clear advantage of this operational point of view is that information theory, which gauges the "amount of information" in terms of the alternative messages that might have been sent but were not, can be applied directly to speech communication between human beings. The human organism is both a generator and a receiver of speech sounds. These sounds are acoustic patterns that vary in time, intensity, and frequency, and these patterns are categorized into the phonetic elements of speech according to certain basic invariances about which we still have much to learn. In speech communication these phonetic elements are the transmitted "messages," and the listener's task at each moment is to discriminate which of the several alternative elements was actually sent.

Wait, I thought that the transmitted message was that observable "glow of understanding"?

Applying information theory to human communication (via speech and otherwise) is a fine idea; and I'm entirely in favor of using "observables" to infer the nature of hidden structures. But the idea that any received message must be operationalized in terms of an immediate behavioral response is idiotic, and is in no way was implied by  Claude Shannon's 1948  "A Mathematical Theory of Communication", as schematized in his famous figure:

How could such transparent nonsense have been promoted by Stanley Smith Stevens, who created and led the Harvard Psycho-Acoustic Laboratory, which produced so much good research and so many eminent researchers? It seems to be an example of a how a socially-sanctioned belief system can lead smart people to endorse and promote  stupid ideas. It's daunting to think that all of us are probably guilty of similar idiocies, seen from the perspective of 2090 or so.

There's a lot more to say about that 1950 "Speech Communication Conference", and even more about the subsequent seven decades of academic takes on "communication" — but sufficient unto the day…



  1. Philip Taylor said,

    August 23, 2018 @ 2:52 pm

    Mark, it is not clear to me why you believe (and assert) that "the transmitted message was that observable "glow of understanding". The 'observable "glow of understanding"' is surely the "discriminatory response" of the reader, is it not ?

  2. Jerry Friedman said,

    August 23, 2018 @ 4:24 pm

    "Appealing to behavioral operation as the test for the presence or absence of communication" makes a lot more sense to me than "Communication is the discriminatory response of an organism to a stimulus." (Emphasis added.)

    For greater overlap with the usual definition of communication, Stevens might have said something about the source, as Shannon did. As far as I can tell, if a hen sees an insect or a corn kernel and pecks at it, communication has happened.

    Can the glow of understanding be facial expressions and such?

  3. Gregory Kusnick said,

    August 23, 2018 @ 5:51 pm

    Conversely, if I read and comprehend the message of a speed limit sign, but fail to slow down for it, communication has not happened. Or is talk of comprehension taboo under this behaviorist definition?

    [(myl) As Stevens said, "If the stimulus is ignored by the organism, there has been no communication. […] [W]e explicitly forsake all concern with abstracted meanings, significations, and the like". ]

  4. Philip Taylor said,

    August 23, 2018 @ 6:03 pm

    Gregory K — I'm not convinced that "failing to slow down" implies that the stimulus (the speed limit sign) has been ignored by the organism (yourself). I don't think that you have ignored the sign, but rather that you have chosen to ignore the speed limit that the sign has communicated to you. Not intended as nit-picking but rather seeking to differentiate between ignoring a stimulus and ignoring the data that the stimulus communicates (or seeks to communicate).

    [(myl) For Stevens, there is no such thing as "the data that the stimulus communicates (or seeks to communicate)". That would be part of the "abstracted meanings, significations, and the like", which he proposes to forsake.]

  5. Andrew Usher said,

    August 23, 2018 @ 10:02 pm

    Strict behaviorism applied to humans results in nonsense, of course, I don't think you'd find much dissent anymore from that. But the following of this post to the Joos one makes one suspect at least for a moment that this, too, could be an elaborate joke. If only it had been.

    By the way, the reason the word communication 'means many things to many people' is that it is such a broad concept and it's only natural for a broad concept to be specialised to the matter at hand; it's not normal to want to talk about 'communication' in the most general sense, though I accept that it can possibly be enlightening to do so.

    k_over_hbarc at

  6. Emily said,

    August 23, 2018 @ 10:15 pm

    I assume that the "glow of understanding" refers to facial expressions and gestures. Such things would be observable, though of course they aren't necessarily much help in narrowing down what was communicated.

    As an aside: I think "modem treatise" is an OCR error due to keming, though it kind of fits the context.

  7. Philip Taylor said,

    August 24, 2018 @ 3:26 am

    To pursue the speeding sign analogy a little further, I very much suspect that even if a driver manifests no overt signs of reacting to a speed limit sign (i.e., he continues to speed), he nonetheless does react to the sign by (for example) making frequent use of his rear-view mirror to check for police cars and the like, and by checking for evidence of roadside/gantry cameras. So I would assert that such a driver has responded to the speed limit sign (and thus that communication has taken place), even if his response is not that which the sign was intended to elicit.

  8. Ursa Major said,

    August 24, 2018 @ 4:57 am

    To go even further with the speed limit sign example: even if the driver does not display the behavioral changes that Philip Taylor suggests and continues driving exactly as if the sign wasn't there, if they mention the speed limit to their colleague the following day has the sign now succesfully communicated?

    But this is actually just a case of not observing closely enough… if the driver "reads and comprehends the message of a speed limit sign" then a sensitive-enough brain scanner should detect the moment of comprehension. Is that response discriminatory? I would say that the driver has chosen to process the fact of the speed limit in way that they haven't for, say, the number of branches on the tree next to the sign.

  9. Ray said,

    August 24, 2018 @ 8:28 am

    "The message that gets no response is not a communication." <– this is hard to grasp. because even when people ignore you, they're communicating something. (this is why it's funny when people ignore you. they're not, really!)

  10. KeithB said,

    August 24, 2018 @ 8:53 am

    From the Far Side:
    Henpecking amoeba wife to amoeba husband:
    "Stimulus, response! Stimulus response! Don't you ever *think*?

    [(myl) Nice.


  11. rcalmy said,

    August 24, 2018 @ 10:26 am

    There seems to also be a problem on the opposite end of the spectrum from the oddness of no response = no communication. The way it's worded, aren't we forced to judge that any stimulus that produces a response is communication? It started to rain while I was walking the dog this morning. In response to the stimulus of being rained on, I turned back early. Do I conclude that the rain communicated with me?

    [(myl) In the ordinary-language sense of the word, obviously not. But there's a (productive) strand of research in the psychology of perception that frames perception as decoding "messages" from the environment, in information-theoretic terms.]

  12. Vulcan With a Mullet said,

    August 24, 2018 @ 10:29 am

    It seems to me that if I see a sign, comprehend it, and ignore it, there has undoubtedly been a communication… by the intuitive definition, anyway.

    Compare this to my overhearing a conversation in a language I don't know. I can see this both ways… there is no communication of ideas, but there may be a clear communication of emotion and tone…

    Whereas if I hear an electrical beeping in a strange pattern coming from a robot, there is truly no communication.

    Seems like a spectrum to me.

  13. Gregory Kusnick said,

    August 24, 2018 @ 12:36 pm

    Ursa Major: With a sensitive-enough brain scanner, any perceptible stimulus will yield a detectable neurological signal. So that can't be what Stevens means by a discriminatory response (leaving aside the point that behaviorists consider any discussion of brain states to be anathema).

    rcalmy: My flip-side example was going to be swerving to avoid a rock in the road. Is that a successful act of communication by the rock?

  14. Margaret Wilson said,

    August 24, 2018 @ 2:26 pm

    It doesn't just happen with socially-sanctioned belief systems. This is a broader phenomena, wherein people who are experts in Very Difficult Fields make fools of themselves as soon as they step outside their area, because they assume that their mastery of Very Difficult Field must mean that they are smart about All The Things.

  15. JPL said,

    August 24, 2018 @ 6:46 pm

    Wow! Bygone age indeed! It would be interesting to figure out just where things have gone wrong here. It's the age of "Logical Positivism" (not using the revisional "Logical empiricism" here), Machian phenomenalism, confusion of the role of evidence for the role of object of study, the abstemious moralistic European stricture to say only what is incontrovertibly true ("Red patch here now", etc.) in the sense of "correspondence with the facts", and so forth. The effects of these conventional attitudes are still with us in some corners of some academic fields. For philosophers: Why did you go along with these tendencies (which eventually led to Tarski, Davidson, etc.) and ignore, e.g., Ernst Cassirer and the Marburg neo-Kantian approach? Even Chomsky joined in the proscription of meaning from the list of respectable objects of study, and still most semanticists stick to the world of formal logic. Does nobody want to try to answer Putnam's question, "How does language hook on to the world?" They mentioned hens: it's sad that they would never understand their clucking with those attitudes.

  16. AntC said,

    August 25, 2018 @ 12:13 am

    It's the age of "Logical Positivism"

    1950. Hmm? Gilbert Ryle's 'Concept of Mind' published 1949.

    I'd say "Logical Positivism" was on its last legs by then — amongst Philosophers.

    A form of LP persisted for some time after amongst (some) scientists/research communities in Karl Popper's (mostly bogus) characterisation of pure science. Its main proponents were exactly those in the disciplines Popper characterised as development rather than research. Perhaps linguistics was still trying to establish its bona fides as a science? I guess that's what was driving Skinner.

  17. Mark Young said,

    August 25, 2018 @ 9:44 pm

    ] Are we really supposed to believe that some people's "glow of understanding" on reading about information theory is an "observable phenomenon"[

    I don't think we are supposed to believe that. The author wrote:

    ]] At a different extreme it [his definition] includes the modem treatise on information theory, which some people seem to read and respond to with a glow of understanding.[[

    I would take the author to be saying that the fact that his definition captures this treatise (that so many find enlightening) is a point in its favour. The enlightenment/glow of understanding that the treatise engenders (communicated, presumably, by behavioural changes) is transferred to his own definition of which the treatise is a special case.

    So his position may well be nonsense, but that particular nonsense is not part of his position.

  18. JPL said,

    August 27, 2018 @ 3:00 am


    I was wondering, in that comment, where the intellectual values that seemed to be restricting some people in this era came from. The question is, why did Stevens feel he had to "explicitly forsake all concern with abstracted meanings, significations, and the like" and "stick to observable phenomena" in order to be affirmed as a conventionally respectable scientist? Here is evidence of a "socially sanctioned belief system" blocking a presumably intelligent person's curiosity. What has happened to his ability to wonder and struggle to understand a puzzling phenomenon and figure out why it is the way it is? He should have taken as an example what humans call a discussion of information theory, but which he would have to take as an endless emission of meaningless noises that all seems so pointless. The message sent by a speaker is regarded (but not consciously) as referentially identical to the phonetic elements that are describable by the phoneticist in physical terms, and the listener's task is to recognize which acoustic signals were produced, and that's it. We stop there. We've somehow managed to emancipate ourselves a little bit from these conventions, but I'm not sure we're completely free of them yet. Q: Does a language system, described by the linguist with a grammar, include, in addition to a system of morphosyntactic rules for (presumably) the expression of meaning, a unified system of rules governing the activity of the meaning expressed? How in fact is a specific meaning, as opposed to a slightly different meaning, possible? One way of expressing a possible idea that probably sounds way out of bounds.

  19. Peter Gerdes said,

    August 28, 2018 @ 2:45 am

    First, the parts you quoted never defined anything about meaning in terms of a discriminatory response. They merely claimed that a communication occurs if and only if some stimulus results in some difference in behavior.

    And where did it ever say *immediate* discriminatory response? Seems pretty clear to me that they want to call anything which results in the organism taking different actions over it's entire future life a communication. I'd prefer the definition to be a bit more expansive and define anything that *could* produce a behavioral difference in some potential future circumstance a communication but the definition they give doesn't seem to be bad.

    Seems to me this is a perfectly legitimate, if underwhelming, definition serving the purpose of differentiating between what we might intuitively call observed and unobserved aspects of the environment without having to bog down in questions about whether a particular stimulus (e.g. a subliminal message introduced in a movie) was truly 'observed' if it made a difference in early parts of our visual processing system but was filtered out before it made it to higher levels or was saved in any fashion.

RSS feed for comments on this post