i think everyone is a scientist: the poverty of stimulus argument

There is a disconnect between how some social scientists see themselves versus how they see their subjects.  Scientists theorize about the world — they develop hypotheses, models, they reason, imagine, simulate, then test and revise, etc — and regular folks, well, learn more myopically via observation and experience. Behaviorism of course represented an extreme case of the latter – a stimulus-driven, passive view of human behavior.

But I’ll go on a limb and say that I think that the “scientist model” is a far better conception of all human activity.  Everyday living and interaction is scientific activity of a sort: we have models of the world that we constantly update and revise.  Importantly, these models have an a priori nature, decoupled from experience.  Does experience matter?  Sure.  But, I think the a priori factors matter just as much, even more.  How one conceptualizes the a priori depends on one’s field and purposes, but it includes the following types of things – human nature, choice, reason, imagination, intention, conjectures, hypotheses and theories and so forth.

Readers will of course recognize the above dichotomy as the rationalism versus empiricism debate: reason versus experience.  Empiricism, very often, looks deceptively scientific.  After all, it’s easy to count things that we can observe.  Experience and history are master mechanisms behind gobs of theories — tracing, counting what happened in the past appears scientific.  In some cases it is.   But, the stuff that we observe and perceive is heavily theory-laden (no, not in that sense), and observations and perceptions might simply be epiphenomena of a priori “stuff.”  And, experience might simply “trigger” rather than cause outcomes.  Furthermore, experience and history are only one of many, possible worlds.

The “poverty of stimulus” argument relates to this.  Varieties of the poverty of stimulus argument show up in developmental psychology, linguistics, philosophy, ethology and other areas.  In short, the upshot of the poverty of stimulus argument is that outputs and capabilities manifest by organisms far outstrip inputs such as experiences and stimuli. The work on infants, by folks like Elizabeth Spelke and Alison Gopnik, highlights this point: children have clear, a priori conceptions of their surroundings.  Wilhelm Von Humboldt’s notion of language capabilities as the “infinite use of finite means” relates to the poverty of stimulus argument.  Some varieties of decision-making models (depending on what types of “priors” they allow) also fit.  Ned Block’s “productivity argument” fits into this.  As does, perhaps, Charles Peirce’s notion of “abduction.” Etc.

The above discussion of course is a very Chomskyan view of human nature and science.  But, this tradition goes back much further (well, to Plato).  In my mind, one of the best, historical primers on some of these issues is Chomsky’s Cartesian Linguistics: A Chapter in the History of Rationalist Thought (be sure to get the 2003 edition, with McGilvray’s excellent introduction).  A very, very under-rated book.

Overall — I’ll go out on a limb, again (no one reads the last paragraph of loose, jargon-laden rants/posts like this anyways) — I don’t think the social sciences have come to terms with the scientific problems associated with experience-heavy arguments and the crucial importance of the a priori (however conceived).  I think there are lots of research opportunities in this space.

About these ads

Written by teppo

March 18, 2011 at 7:06 am

15 Responses

Subscribe to comments with RSS.

  1. Nice post Teppo.

    I think the distinction between normal human experience and science is that the latter consists of a set of precise methods that can be used to fine tune our sense of observation and ability to make inferences. Science helps humans cut through the natural perceptual and cognitive biases that lead us to commonly make incorrect inferences.

    I think a great example of how science helps to improve inference-making is to look at the Sabermetric revolution in baseball. Sabermetrics is the use of statistical analysis to evaluate (and value) player performance. Prior to the use of statistical analysis, professional scouts were the accepted experts on player quality. They relied on their deep experience and familiarity with the game to assess how young players might fit into a major league system. They were often wrong, focusing on qualities that they personally valued (e.g., shoulder-to-hip ratio) and had come to associate with success, but that had a very low correlation with player achievement in the majors. Once a few teams started adopting statistical analysis to evaluate players, those teams quickly developed a competitive advantage and were able to identify undervalued players that the scouts, with their immense experience and conceptual models, had missed. Of course, it only took a few years for all of the teams to catch on to what the sabermetricians were doing, and now every team has a group of stats nerds who do nothing but run models and evaluate talent. It’s completely changed the way general managers evaluate and promote players.

    So, the point is that even a simple scientific method, like statistical analysis, can help overcome natural human biases that have led generations of humans to make incorrect inferences. That’s the real benefit of the scientific revolution – it greatly improves our ability to observe and make correct inferences. It’s a natural human process (which I think is your point) but we’ve just figured out how to make it work better.

    brayden king

    March 18, 2011 at 2:17 pm

  2. I have often described entrepreneurs as offering and revising conjectures in the Popperian sense; they put forth a business plan, resource combination, etc. and examine it against the data of market reality, after which they revise or reject (according to profit-and-loss signals).

    Peter Klein

    March 18, 2011 at 2:53 pm

  3. Welcome to Chapter 1 of Pareto’s Treatise on General Sociology.


    March 18, 2011 at 6:53 pm

  4. Brayden: Right, clearly there is some difference between the science of scientists versus the science of everyday life. Though, some scientific approaches implicitly create a mega-gap that I think is untenable (as if we’re dealing with two species). Naturally, some folks would also lump science itself into an observation-dependent affair (Einstein, Goedel, Mach, Popper and other Vienna folks of course debated these issues, and the debates continue). Here’s Peirce on that point: “man’s mind has a natural adaptation to imagining correct theories of some kinds…If man had not the gift of a mind adapted to his requirements, he could not have acquired any knowledge.”

    Peter: Todd and I have also found the notion of “entrepreneurs as theorists”/scientists helpful.

    James: Sure — there are certainly strands (and hints) of this intuition in social theory: Pareto, Coleman’s sociology has some elements of this (also, Boudon and choice-type folks), some of Simmel, some strands of economics (e.g., Austrian conceptions and others), etc. Though, I think that a bulk of work is directly antithetical. But, naturally the whole “poverty of stimulus” intuition is also under debate (with strong proponents on both sides).


    March 18, 2011 at 7:20 pm

  5. Sorry, I said that in a bit of a snarky way. It’s been a long morning.

    Though seriously, Pareto’s explicit focus on sociology as the study of theories – both logical and non-logical – as they affect human action seems to speak really strongly to your blog post.


    March 18, 2011 at 7:50 pm

  6. Teppo – you could also make the analogy work in the other direction. Scientists are like entrepreneurs. We rarely get the theory right on our own (and rarely on the first try) – in fact, the likelihood of failure for any given scientific enterprise is pretty high – but in the long run, we collectively arrive at the right solutions (e.g., the move to Copernican theory would not have happened without hundreds of other little discoveries and failures along the way).

    brayden king

    March 18, 2011 at 8:19 pm

  7. Are you familiar with the work of Paul Thagard?

    Ruchira Datta

    March 18, 2011 at 8:42 pm

  8. Ruchira: Yes, I’ve read some his stuff (role of analogies, association, similarity, etc, with Holyoak), though its been a while. The approach is a variant of connectionism, I guess (big debates there as well, w/ folks like Fodor and others highly skeptical). With the emphasis on similarity, connectionist approaches seem to share much with empiricism. But, here I’m getting into -ism soup – certainly there are many variants within and in between these positions (and varying levels of confidence in our abilities to actually understand the underlying processes).


    March 18, 2011 at 11:12 pm

  9. you can get an introduction to the kind of developmental psychology Teppo is describing is shown in the documentary The Baby Human or with lecture 5 of Paul Bloom’s intro to psych course. you know, that or journal articles and books.


    March 19, 2011 at 2:04 pm

  10. The origins of this blog post…this stuff is top-of-mind for me right now as Nicolai and I are involved in a (friendly) debate/discussion (with Sid Winter, Brian Pentland, Geoff Hodgson and Thorbjoern Knudsen) that directly implicates these issues. In short, we think the emphasis on experience and history is misplaced and problematic, and the individuals-as-theorists intuition needs to be taken seriously in extant org theory and strategy (and social theory more broadly).

    Here’s our piece on this (and Sid’s and other’s papers are also posted online) – The endogenous origins of experience, routines and organizational capabilities: the poverty of stimulus.


    March 19, 2011 at 5:25 pm

  11. The arguably leading approach in categorization research is “theory theory” (aka knowledge-based approach). To my knowledge the cognitive psychology research has since mid 90s started very much to look at the role of prior knowledge in shaping psychological processes. This is supposedly a big change because most laboratory research focuses on controllable stimuli and thereby ignore (/actively seek to make irrelevant) what people know.

    That leaves much use of psychology within strategic management looking very outdated.

    I would imagine that everyone always knew that knowledge and theories matters a lot (Strang & Meyer 1993 paper on the role of theorizations in diffusion of practices; Meyer et al. 1998 paper on world systems also acknowledges the importance of superior rational theories in cultural dominance). The problem is that the preference for quantitative methods and the non-existing attention span of reviewers call for papers to be simple. Taking into account theories held by actors is simply not possible in a quantitative paper. It is also far more easier to make negative contributions (“theories matter so the prior research is wrong”) than to make positive contribution (finding out new regularities related to theory-use by reserach subjects).

    Thus, the institutional structure of academic work forces social scientists into research traditions ignorant of some really important phenomena.

    Behavioral economics is a good example of enthusiasm created by quantifiable and researchable topics. Kahneman’s 2003 article begins by acknowledging that reflective knowledge use matters, but notes that the biases in intuitive reasoning are still really important in decision making. We need to wait until someone figures out how to quantify how prior knowledge structures (or theories) influence economic decision making.


    March 20, 2011 at 7:48 am

  12. Teppo, this got me to pick up a book on semantic cognition I was reading. One reference cited in the book that might be of interest is Alison Gopnik (

    [...] she formulated the “theory theory”, the idea that children’s learn in the same way that scientists do.

    So as you stipulated, everybody is a scientist — including small kids.


    March 20, 2011 at 7:50 pm

  13. A few, additional resources:

    Chomsky’s 2002 Berkeley lecture –

    I like (and frequently cite) this 1992 piece by Spelke et al –


    March 20, 2011 at 11:31 pm

  14. [...] can make a strong case for trying to make these choices more like a scientist does. Here’s an [...]

  15. [...] can make a strong case for trying to make these choices more like a scientist does. Here’s an [...]

Comments are closed.


Get every new post delivered to your Inbox.

Join 970 other followers

%d bloggers like this: