Fast lemons and intuitive beliefs

Is a lemon fast or slow?

Which one is brighter: the sound of a violin or the sound of a trombone?

Got the answer?

Without any apparent reason, you believe that lemons are fast and violins sound brighter than trombones. These beliefs happen to be shared by most humans, from an early age and cross-culturally. Now, where on earth did we get them from? Most of the earlier studies conducted since Edward Sapir focused on “sound symbolism”, i.e. the associations between sounds and meanings, but not directly on the associations between sensory dimensions – like brightness, pitch, size, etc. Even if you don't know anything about french names for birds, I could ask you whether you think that a pipit is a small or a big bird – and you would, without any apparent reason, judge that it must be a small one. It's certainly because, as a large majority of humans across cultures and linguistic groups, you think that the sound /i/ is smaller and brighter than the sound /a/, diminutive words, or names of small birds and fishes are much more likely to contain /i/ than /a/.


Rules have exceptions: Which is faster — a mallard or a kiwi?

Believing in things that don't really make sense and without any apparent reason seems, in that respect, not specific to the religious and spiritual domain. But are beliefs in the fastness of lemons, and in bright violins, of the same kind as beliefs in the Holy Trinity or in the spirits of the trees?



Believing is not a simple and unified thing. This anthropological fact is often minimized by philosophers, but fortunately not on this blog (see the debate between Maurice Bloch and Dan Sperber, here and here). Still, when invited to think about varieties of beliefs, most will agree that it comes from two things. The first and obvious one is the variety of contents. Believing that 2+2 equal 4 is different from believing that it will rain tomorrow, and believing that Cicero was a great orator may be different from believing that Tully was a great orator. The second kind of difference comes from our attitudes toward these contents – broadly speaking, the degree of confidence we have toward them. You are absolutely certain that 2+2 = 4, but only pretty confident that it will rain tomorrow.



Reflective vs. intuitive: Where should we look for the difference?

A further distinction illuminates the varieties of beliefs, stressing the difference between ways in which we come to hold these contents – e.g. in a intuitive or reflective way. You can have the intuition that it is wrong to push someone under a train to save five innocents, but come to reflectively believe that this is the right thing to do. This seems close to the previous difference in attitude toward the contents: we feel justified in holding the second belief, but not the first. The two distinctions should nonetheless be kept separate. One can be very confident in one’s intuitions, or on the contrary be suspicious of them and refuse to act according to them. The same goes for reflective judgements. Saying that someone is “very intuitive” is here misleading, and underscores something about attitudes: it doesn’t mean that she has more intuitions than others, but that she trusts and follows her intuitions more easily or often.

The distinction between intuitive and reflective beliefs often goes with dual-systems theories. It is drawn at the sub-personal level in terms of kinds of processes: intuitions result from processes (system 1) which are associative, unconscious, effortless, heuristic, whereas reflective beliefs come from rule-based, conscious, effortful, analytic, and rational processes (system 2).

Yet, as Dan suggests in his post, there is a firmer manifest difference between intuitive and reflective beliefs – a difference that is consciously experienced and expressed in behaviour. Our beliefs come in different ways – intuitions are felt to be “mere knowledge of plain facts”, they come as obvious and somehow isolated, whereas reflective beliefs are felt to be held “for a good reason” (be it an argument or trust in the source) and to make sense within a more general set of beliefs, which one feels are lying in the background, ready to become explicit. This difference is also (and perhaps more clearly) manifested in the ways these beliefs are expressed or discussed: intuitions are stated but can’t lead to further elaboration, whereas reflective beliefs are discussed, argued for and criticized.

A first question here is whether the most important or basic difference between intuitive and reflective beliefs should be made at the manifest level. Perhaps this is a bit fragile (qua an introspective difference) or close to old-fashioned behaviorism (qua observed in practice). But arguably, this difference explains why and how we look for potential underlying differences in processing.


Looking for a distinction in contents: Are some contents specific to intuitions, or reflective judgments?

A more robust way to draw the difference resorts to something more akin to the difference in contents, than ways of being manifested. Following this line means finding contents that are specific to intuitions or to reflective judgments. In their discussion, Dan and Maurice argued about such a difference. Let’s grant that beliefs contents vary in terms of being more or less determinate. As adults, our belief that 2 plus 2 equal four is fully determinate, in the way it wasn’t when we were 5 years old, and in a way that the dogma of the Holy Trinity isn’t. This is, I guess, the distinction that Sperber tries to capture in terms of propositional and semi-propositional content. Terminology doesn’t matter here. What matters is that the distinction doesn’t reduce to a difference in attitude or commitment (our readiness to doubt or wonder about semi-propositional beliefs, but not about propositional beliefs) but remains about contents. Accordingly, a believed content is fully determined (“propositional”) if the believer knows its truth-values over the range of all epistemically relevant worlds, vague if she knows them for some or most epistemically relevant worlds, and indeterminate (only “semi-propositional”) if she doesn’t know anything about its truth-value in any relevant epistemic world.

Let’s go back to intuitive vs. reflective. One can think (like Dan Sperber) that indeterminate (semi-propositional) contents are specific to reflective judgments. Our reason for accepting these obscure contents can’t come (by definition) from our evaluating their truth, and therefore only comes from us trusting the people whom we get them from. “Semi-propositional” contents are always social – they result from communication (linguistic or non-linguistic) and are held for reasons like trust, deference, etc. It is because a child is told that two plus two equal four or that there is something like the Holy trinity that he comes to believe it. He has no clue on what makes it true in the “relevant epistemic worlds”.

While I agree that indeterminate contents can’t come from an evaluation of truth-value, i don’t think that they necessarily come from others. There is much room for indeterminacy coming from our own cognitive activities. Let's reconsider the previous intuitive judgments, about lemons being fast and violins bright. Or take better known cases of sound-shape correspondences. Watch for instance these two shapes:

One is called Bouba, and the other is called Kiki. Which one is which? As almost all of the subjects, you’ll say that Kiki is the star-like one and Bouba the rounded one (Ramachandran, & Hubbard (2001)). Or again: which of the following shapes is called Mil and which is called Mal (Sapir, 1929)?



You were able to answer these questions and judge that Mil is (or must be ) the smaller circle, yet have no idea of the reasons for which you made it. It seems obvious, and there is nothing more you can connect it to – like a good reason to say this. Judgments about cross-modal correspondances (e.g. holding between two apparently unrelated sensory dimensions or features) are being studied by cognitive scientists (see Spence, 2011 for a review). This goes back to the initial example – we intuitively know the answer to an apparently insane question like “are lemons fast or slow?”.


The varieties of intuitions

There are many intriguing aspects of these correspondences and the judgments they lead to. For current purposes, the point is that these are relevantly called beliefs: people commit themselves, at least to a certain degree, to their truth, and most of them can be entertained in an explicit way. Now the contents of these beliefs certainly fail to be determined in any of the previous senses. There is simply no way to make sense of what it is for the content “this shape must be called Bouba”, “this shape can’t be called Kiki” or “lemons are fast” to be true. Even what underlies the term or concept of correspondence, congruency or going-togetherness which structures these intuitions that “these two things must go together”, “Bouba goes more appropriately with that shape” seems to be beyond proper understanding and definition.

In a nutshell : this seems to be a very good case of intuitive indeterminate (non-propositional) content. Moreover, we cannot simply explain (at least by now) how these correspondences could come from things we perceive or sense. Being effortless, made without attention or awareness of reasons, based on broadly associative ways, these beliefs are certainly intuitive.

I don’t know what it implies for other “symbolic” beliefs – noticeably closer to the religious domain. Perhaps it gives us ways to think about some of them also being grounded in intuitions of this sort (I doubt that this would work with the Holy Trinity, but perhaps with other contents? Any idea of religious beliefs that would work along the lines of correspondences?).

More interestingly, it challenges us to think about another difference in kinds of contents for the intuitive vs. reflective distinction. Failing to be propositional or determinate is not specific to reflective judgments, or so I think these cross-modal cases show. As a further challenge, it also invites us to focus not just on the varieties of beliefs but on the varieties of intuitions. My suggestion here is that intuitions whose contents are somehow grounded in perceptual processing (but not perceived per se), like cross-modal correspondences, may represent a very distinctive subset of intuitions, contrasting with moral or linguistic ones, for instance.


Some references

On intuitive vs. reflective beliefs

Sperber, Dan (1997) Intuitive and reflective beliefs, Mind and Language 12 : 67-83 (link)

On sound symbolism

Nuckolls, J. (2003). The case for sound symbolism. Annual Review of Anthropology, 28: 225–252. (link)

Ramachandran, V., & Hubbard, E. (2001). Synaesthesia: A window into perception, thought and language. Journal of Consciousness Studies, 8: 3-34. (link)

Sapir, E. (1929). A study in phonetic symbolism. Journal of Experimental Psychology, 12: 225-239. (link)

Spence, C. (2011). Crossmodal correspondences: A tutorial review. Attention, Perception, & Psychophysics, 73: 971-995. (link)

On dual processes

Evans, J. St. B. T. (2008). Dual-processing accounts of reasoning, judgment, and social cognition. Annual Review of Psychology, 59: 255–278.

Keren, G., & Schul, Y. (2009). Two is not always better than one: A critical evaluation of two-system theories. Perspectives on Psychological Science, 4: 500–533.

Kruglanski, A.W. & Gigerenzer, G. (2011) Intuitive and Deliberate Judgments Are Based on Common Principles, Psychological Review, 118: 97-109.




  • Olivier Morin
    Olivier Morin 6 June 2011 (12:18)

    Thank you Ophelia! I loved this post, but I am reluctant to endorse your view. You claim that our intuitions about phonetic symbolism are not propositional in Sperber’s sense. “There is simply no way to make sense of them” as they are “beyond any proper understanding or definition”. Well… not so fast. The psychologists that you cite find it quite easy to define and understand phonetic symbolism: certain sounds, shapes or colors seem to go together with stimuli from other modalities because they evoke certain synaesthetic associations. This is sketchy, but not mysterious. Of course, most people just feel the impact of those synaesthetic associative mechanisms, not knowing what they might consist in. But do they need to? A propositional intuition is not an intuition that is fully explained or justified (otherwise, come to think of it, vision would not be propositional). A propositional intuition simply has to be “understood well enough by the believer for ordinary purposes: she knows well enough what the belief presupposes and entails in ordinary situations” (Sperber, from [url=]the post you cited[/url]). When they choose a name for a child, a character in a novel, or a yoghurt brand, ordinary people use their ‘symbolic’ intuitions quite skillfully. They appear to know how these intuitions work, at least well enough to use them in daily life. Of course, these intuitions are seldom fully justified or explained, but most of our intuitions never are. Our knowledge of beauty, warmth or acidity hardly lends itself to rational justification, but that does not make it semi-propositional. Or does it?

  • Paulo Sousa 7 June 2011 (21:48)

    Thanks for this interesting post, Ophelia. I’m not sure whether the basic premise in your argument is warranted–i.e. that people have a belief whose content is [i]a lemon is fast[/i], for example. The fact that people have strong intuitions about the ‘right answer’ to an alternative question like “is a lemon fast or slow?” does not seem to be enough evidence that people have a belief with such a content. In other words, does this type of ‘forced choice between two alternatives’ indicate that the content [i]a lemon is fast[/i] plays the role of a belief in one’s mind? Perhaps, it just indicates that, given the alternative, this is obviously the best answer.

  • Helen De Cruz
    Helen De Cruz 9 June 2011 (12:12)

    Ophelia, the beliefs you mention in this text are probably best described as ‘tacit beliefs’. Tacit beliefs are not stored in a representational, mental format, but rather, are spontaneously formed or assented to in particular situations. Dennett gives somewhere the example that most of us would spontaneously agree that ‘zebras don’t wear overcoats in the wild’, although few have had this stored as a mental representation. More formally phrased, we can say that an agent S [i]occurrently[/i] believes p iff she has a mental representation of p. By contrast, S [i]tacitly[/i] believes p iff S spontaneously and reliably forms p in a specific situation or context, but prior to this, S did not have p stored as a mental representation. Some philosophers of mind are unhappy with tacit beliefs, and would like to eliminate them only together, focusing only on occurrent beliefs (e.g., Manfredi, and if I read Paolo’s comment correctly, he would also endorse this). However, your examples suggest that tacit beliefs may play some interesting role in our mental lives, maybe we shouldn’t eliminate them just yet.

  • Nicolas Baumard 9 June 2011 (19:54)

    Thanks Ophelia for this very refreshing post! (I keep thinking about the speed of lemons, which enlivens my day.) I think the case of intuitive association is a very useful case in point to understand the distinction between intuitions and reflections, particularly in the domain of religion. Take the case of immanent justice. Jack trips up and breaks his leg. I recall that Jack has cheated on his partner. Because these two events are quite salient in my mind and because they seem roughly proportionate, I have an intuition (an association) Jach’s misfortune compensate his misdeed and restore fairness. Now it is just an intuition. I do not have any theory about whether Jack has been really punished, owhether there is a causal relation between his misfortune and his misdeed, or whether such things as God, Destiny or Fate exist. I just have the intuition that it is ‘only fair’ that he broke his leg. I do not really believe in immanent justice. I just can’t help thinking that Jack got what he deserved. To sum up, the focus on intuitions or associations reveals that they cannot, in themselves, explain the existence of explicit beliefs according to which misdeeds and misfortunes are thought to be causally related (through supernatural justice, punishing gods, vengeful ancestors, etc.). Intuitions probably help to religious beliefs to spread and enjoy cultural success but they do not lead directly to religious beliefs. More on intuitions and reflections in immanent justice [url=]here[/url].

  • Paulo Sousa 10 June 2011 (18:32)

    Salut Nicolas. Although I’m quite convinced by your claims about our modular intuitions of fairness (I’ve read your two recent articles, with Andre and Sperber, on the topic: quite interesting stuff), I don’t think your additional comments about association are really relevant. a) In relation to the intuition [i]it is only fair that he broke his leg[/i], which is an intuitive conclusion coming from the computations of the fairness module, to say that this is an association is misleading because the inferential processes that lead to the intuitive conclusion are much more complex than the computation of simple associations (e.g. it involves a bunch of modular assumptions like the assumption of proportionality). You may think that the usage of “association” here is appropriate because “association” normally suggests something weaker than causation, and you deny (rightly so, I think) that these modular computations involve inferences about causal processes (e.g. the inference that he broke his leg because he cheated). But in this case this usage just indicates that these computations do not involve such causal inferences. b) In relation to the intuition [i]considering that it is fair that he broke his leg is linked to considering that he cheated[/i], to say that this is just an association may be misleading as well. This link is rather about why one reached the intuitive conclusion delineated in “a”. In other words, it is rather a somewhat good intuition about the modular computations of the fairness module—although one does not have conscious access to the inferential processes of the fairness module, one may well have some good introspective evidence that thinking that it is only fair that he broke his leg comes from thinking that he cheated. But now this link is not just an association in the weak sense that excludes causality because it is about a causal relation, though about a causal relation in one’s thinking. Perhaps, it is this meta-causal element that contributes to reflective theories to jump in—to postulate an objective causal relation between cheating and breaking a leg is quite compatible with one’s introspective intuitions about one’s mental processes.

  • Nicolas Baumard 17 June 2011 (18:38)

    Salvação Paulo, I totally agree with you! On a), I just wanted to pointed out that association were good examples of intuitive processes and that their study demonstrates that intuitive processes do not directly explain reflective and religious beliefs. But I agree with you that the intuition of immanent justice may be rooted in more complex processes than simple associations. As for b), there is a misunderstanding. Like you, I do not think that the belief that one has been punished is an association. My hunch is that it is influenced by the fact that, in our society, cheaters are typically punished by the penal system. But it is only an hypothesis.

  • Ophelia Deroy 20 June 2011 (10:17)

    Thanks to you all for your comments and reactions….and all the more that they confirm the difficulty to find a status for these “shared replies” about the fastness of lemons, the darkness of low-pitch sounds, etc. The heaviness of colors (yellow is lighter than red, right ?) is also a nice example. I get the legitimacy of your worries and suggestions – but I am not convinced that : – they are “synesthetic associations, intuitive yet propositional (Olivier): first, because they are not synesthetic in any clear sense (it’s not because the sound of trombones triggers an experience of a dark color that you judge that the sounds of trombones are dark – or darker) despite what most people in the literature tend to believe ; second, I am not sure we agree on the notion of “propositional content” (I agree that people are able to express these things linguistically, but this is not enough to say that it fully determines a propositional content, noticeably that there is a clear concept of “going togetherness” here, or that the content “a lemon is fast” is truth-evaluable) – not really believed but forced choices (Paulo) – that’s an important point, indeed. Most of these judgements are triggered in comparative matching tasks (Bouba-Kiki, etc.). This apparently makes them different from most moral intuitions or associations that are triggered one a case by case basis (but I defer here to the experts). Maybe this is the same worry that Helen partly expresses – aren’t these things just ‘contextual’ and an artifact of the question ? Well, the fact is that people show great consistency over time in their replies, so at least they are ‘grounded’ in more than conceptual memory (we agree that if I give a ungrounded reply to a silly question, I am very unlikely to remember it and give the same reply five years later). Tacit beliefs, then ? Again, to call them beliefs would mean that they have propositional content, and this is what I don’t really agree with. – intuitive associations (Nicolas) : I certainly like the comparison you draw (and hope you’ll be able to think about the heaviness of red, now, as a distraction from the fastness of lemons !) and I am glad to see that they are similar worries for moral intuitions. I guess part of the difference (and problem) comes from what is associated. In your examples, the association is between two propositional contents, although there is no (reflective) reason to put them together. In the case of sensory associations, it is not as clear – do we associate conceptual representations of lemons and speed? sensory images? Most of these examples seem to be better explained by a correspondence between sensory dimensions or magnitudes, that are not distinctively grasped at the level of perception. Well, at least that’s what we want to argue – in a forthcoming paper. Perhaps I should also add that lots of these “matchings” show an influence on cross-modal binding – which shows they are not “mere beliefs”, floating above the rest of our cognitive systems. Some of them have also been demonstrated in infants – as well as rats and macaques (although here, the change of methodologies can raise doubts as to whether we are testing for the same thing….) My (dangerous?) idea is that they could dispense us to think about innate conceptions of objects and explain objectual unity at a lesser coast. Which brings us quite far from fast lemons, but close to good problems. Thanks again for your sharp comments !