Opacity tasting with Dan and Maurice
A few weeks ago, Maurice Bloch, Dan Sperber and several others had a debate on the nature of those beliefs (in particular 'religious' beliefs) that cannot easily be made fully explicit, or brought to bear on concrete matters (so-called 'semi-propositional beliefs' - here called 'opaque' beliefs). The question, very roughly, was the following. Can such beliefs be arrived at only through some form of reasoning or of trust? In other words, must they be 'reflective' rather than simply 'intuitive'? Maurice argued that such beliefs can be intuitive, but Dan disagreed (if you missed the first installments of the feuilleton, see here and here). Dan imagined a friendly discussion with Maurice, taking place around a glass of wine, which has now prompted György Gergely to jump in the debate. (O.M.)
Under the mild mental pleasure that the virtual and vicarious consumption of wine in the distinguished company of such authorities on both wine and opaque beliefs (pardon me for the expression) as Dan Sperber and Maurice Bloch induced in me, I feel liberated to raise some – admittedly opaque but for me at least mildly intoxicating – questions concerning Dan’s characterization of the nature of having reflective semi-propositional beliefs.
What it is like to be holding a reflective belief of opaque content?
It seems to me that Maurice (irrespective of the amount or even quality of wine consumed) is taking a phenomenological approach to the subjective experience of holding a belief when he insists on the identity of having what Dan distinguishes as intuitive versus reflective beliefs (when claiming that the Zande believe in cows and witches alike). This is not the relevant issue, however, for Dan for whom the question of ‘how it feels like’ seems orthogonal to how we define it: “I keep telling Maurice that a belief may be wholly familiar and self-evident to a believer and still not be intuitive as I have defined the term”.Dan’s definition of different types of cognitive attitudes to beliefs makes reference to different kinds of inferences and justifications that the believer of reflective beliefs is ready (or feels appropriate) to engage in (in contrast to the believer of an intuitive belief) in so far as reflective beliefs licence justification by deference to authority in the first place.
This characterization together with the proposal that reflective beliefs are “held with awareness of the reasons and often with attention to those reasons” raises two – admittedly psychological and even empirical – questions:
Do reflective beliefs imply or require cognitive access to the memory of the source authority?
In other words: Does Dan’s distinction entail that reflective believers have stored memory representations of, and ready cognitive access to the source authority of the belief (say, memory of the communicative event(s) that involved the revelation of the belief content by the authority (or its representative) to whom justificatory deference can then be made (when asked))? From the developmental point of view this sounds unlikely in so far as young children are known to be notoriously bad in source memory access when it comes to reporting the sources of their beliefs. Would Dan’s view predict that this should hold at least to a lesser degree in the case of the memories of the causal sources of reflective beliefs that would be encoded better or represented in a more accessible format than the sources of intuitive beliefs?
Alternatively, the contents of reflective beliefs could be encoded in a special representational register or format that would specify their justificatory conditions as ‘deferrable to authority’ without the support of having the actual source memory of the historical communicative event of belief-fixation that would specify the source authority to be deferenced. In this case the question is: What input conditions trigger the special kind of ‘representational register of format’ that tags the belief content with the justificatory condition of being ‘deferrable to authority’?
Deference to authority versus deference to common knowledge.
A related issue is that apart from justification by deference to authority many reflective beliefs seem de facto characterized by a licence for justification by deference to common or public knowledge that are shared by other members of the community as well. These two types of justifications are clearly not unrelated, of course, but – at least from the perspective of natural pedagogy theory (Csibra & Gergely, 2009, in press) – it seems reasonable to argue (and some recent evidence is now available to support this view, Egyed, Király, & Gergely, submitted) – that ostensively communicated acts through which opaque belief contents are demonstrated trigger in infants a built-in assumption that the manifested content represents and conveys a belief that is shared by all members of the cultural community (therefore, they are treated from the start as applicable and generalizable to others).
For some types of opaque beliefs at least, when asked the impertinent ‘why do you believe that X’ question it seems natural to defer justification to common knowledge (“well, doesn’t everybody knows this?”) without making reference to any specific source authority such as “this is what my mother, the priest, the shaman, or the school teacher told me” (even though these authorities may well have been the actual ostensive sources of the belief acquisition). Examples of this would be reflective opaque beliefs such as “the word ‘stone’ refers to stones”, “after ‘eeny-meeny’ comes ‘miny-mow’”, , “a cork screw is for opening wine bottles”, and maybe “‘blue’ is the colour of sadness” or even “witches travel on brooms” and “they are wicked”.
Should we consider these cases to be identical to reflective beliefs whose justification involves primarily deference to authority? Are the former just derivatives of the latter? And isn’t there a (theoretical) need to capture a child’s differential cognitive attitude in relation to justificatory deference when it comes to his or her reflective opaque beliefs such as (1) versus (2)?
“there are negative numbers”
“the word ‘stone’ refers to stones”?
It seems more appropriate for a pupil to justify her partially opaque belief in (1) by deference to the authority of the trusted teacher rather than by deference to common knowledge, while (in the case of one’s mother tongue at least) the opposite seems to hold in the case of (2). Or should we consider (2) an intuitive belief fixed through our evolved LAD that maps words onto basic concepts in our innate conceptual repertoire making (2) to be on a par with our intuitive beliefs that are individuated by our perceptual system? And what about our beliefs in the proper function of artefact kinds such as (3) to (6)?
A knife is for cutting.
spoon is for eating soup.
A phone is to talk to others who are far away (and have a phone).
A watch is for showing the right time.
Infants (as young as 10-month-olds, see Futó, Téglás, Csibra, & Gergely, in press) seem ready to form beliefs about the essential functions of novel artefact kinds on the basis of ostensive communicative demonstrations of their functional use. If word-concept relations were to be considered intuitive beliefs fixed by LAD (I really don’t know if Dan or Maurice would consider this option to be far-fetched), should we by analogy also consider that artifact functions are intuitive beliefs fixed by AAD?
Types of opacities, or types of ways we (come to) entertain them?
As an (as yet somewhat opaquely) related further issue I feel that Sperber’s construct of reflectively believing partially opaque beliefs covers two types of cognitive attitudes that may be useful to differentiate.
First type Opacity : Opacity with a dim (guiding) light at the end of the tunnel: an ‘enlightened’ reflective attitude to semi-propositional beliefs that involves a ‘presumption of full propositionality of content’.
It seems that in certain cases one acquires and represents reflective semi-propositional beliefs with an enlightened cognitive attitude of (temporary) credulity in the source of authority. For example, when the relevant authority informs the pupil that there are negative numbers and provides symbols to refer to them, the child will represent and store this – for her currently still largely opaque – semi-propositional belief with a cognitive attitude of trust in the possibility and prospect of rational discovery of its (presumed) fully propositional content. This kind of reflective but partially opaque belief seems to be stored and encoded together with the implicit promise that it is worth exercising one’s critical faculty of exploring coherence conditions to check for compatibility with other intuitive and reflective propositional beliefs that one holds. In other words, in such cases one holds one’s reflective semi-propositional belief with the ‘presumption of full propositionality of content’. This presumption is reflected in the kind of effortful inferential work that the cognitive attitude supports driving the individual towards rational attempts at discovering and understanding well enough the (presumedly) fully propositional content of the reflective belief. Given that the promised cognitive land of full (or good enough) understanding of the belief’s propositional content is successfully reached (as a result of such reflective inferential efforts), the validating context of the reflective belief will become transformed from justification by deference to authority into justification by deference to (well-formed) arguments. As a further move towards expertise (see Sperber (1997) in Mind & Language) the eventual automatization of the knowledge-base and the inferences involved in the justificatory argumentation about the propositional belief will even result in the reflective propositional belief becoming an intuitively held propositional belief.
Second type Opacity : Opacity in the glow of essential darkness: a conservative or fundamentalist cognitive attitude of credulity towards the contents of semi-propositional beliefs.
These are cases when one has the cognitive attitude of ‘wariness’ about the ‘advisability’ of attempting to ‘understand’, i.e., to critically apply one’s rational faculty to modify the partially opaque content of the semi-propositional belief in order to clarify it (in the hope of turning it into a fully propositional belief). A prime case is, of course, provided by ‘religious’ beliefs such as Dan’s (justifiably favourite) example (7):
The Father, the Son, and the Holy Ghost are one.
The cognitive attitude of ‘wariness’ towards attempting to rationally clarify or modify the opaque content of such reflective beliefs seems to suggest that the cognitive attitude towards such belief contents does not involve a ‘presumption of full propositionality of content’. Rather, the attitude reflects the (tacit) understanding that this kind of opacity is essential and/or socially constituted: that it is ‘public or shared opacity’ that should not be fiddled with (as the belief functions only as long as its opacity is shared, reserved, and respected by all). To the degree that this kind of opacity of content is indeed socially constituted and shared, to change and modify its opaque content by any one individual holding the belief would endanger the sharedness (and consequent functionality) of the opaque belief. Such individual attempts at cognitive clarification of this kind of opacity would run the risk of dissolving the glue of shared knowledge (however opaque) that constitutes the content of the reflective belief. In other words, one’s justificatory criterion of deference is one’s (opaque) belief that one’s opaque understanding of the belief is ‘opaque in the same way’ as others’ opaque understanding of it. As if there would be a social contract about the opacity of the shared belief whose functioning is being threatened by enlightened individual attempts at modifying semi-propositional opacity to achieve full propositional clarity.
My question to Dan is how we should capture (representationally as well as developmentally) the nature of these basically different cognitive attitudes towards reflective semi-propositional beliefs. One possibility is to suggest (as I feel Dan would) that there is no fundamental distinction to capture here: at the time of the acquisition of the reflective semi-propositional belief content the child takes an epistemically enlightened attitude always (i.e., initially entertaining the ‘presumption of full propositionality of content’ in all cases).
In this view, this initial ‘enlightened epistemic stance’ can become changed due to two kinds of sources of information: a) by reaching a cognitive understanding that the search for full propositionality is futile because the “arguments about the content are bound to be inconclusive, which may be a reason to avoid them altogether (as do the Zafimaniry)”, and b) by the cognitive impact of the social reactions of the trusted authorities to the (naturally arising?) reflective inquiries of the child, that represent rebuff, scorn, or even punishment and ostracization on the one hand, and arguments of the exigesic sort on the other, that “are in favour of interpretations that are themselves semi-propositional, and hence the exegesis never comes to a conclusion.”
So let me raise my imaginary glass for a refill.
Benson Saler 21 July 2010 (17:46)
Dan Sperber’s explication of “semi-propositional beliefs” involves some subtleties that I have difficulty in grasping. Long ingrained habits of thought, for instance, make it difficult for me to accept the idea that while there are “semi-propositional beliefs,” there are no “semi-propositions” as such. Anyway, that’s my problem, and I’m still working on it. Perhaps I shall awaken one morning shouting “Eureka.” In the meantime, I think that it might prove useful to consider another approach to the problem of beliefs that on first inspection seem to be irrational. I refer to a long-neglected essay in the journal “Man” by the British philosopher David E. Cooper (1975: 238-256). Cooper has published quite a lot on a diversity of subjects (those of you with an interest in gardening might take a look at his book “The Philosophy of Gardens”). His 1975 essay, “Alternative Logic in Primitive Thought,” insofar as I am aware, is largely unknown to those of us with a contemporary interest in cognition and culture. I gave a very brief description of Cooper’s approach in an end-note to a paper that I published on Lucien Lévy-Bruhl. I paste in that note here in the hope that it may be of interest as yet another alternative possibility in our efforts to understand beliefs: 3. Jan Lukasiewicz (1878-1956), a Polish logician, developed a three-valued logic in 1917, the values being “true,” “false,” and “possible.” He built on this work, as did various of his students (e.g., Mordchaj Wajsberg and Jerzy Slupecki), devising multi-valued logics, and so demonstrating that in propositional logic there are viable alternatives to bivalent “standard” logic (Aristotelian logic). Lukasiewicz's work (e.g., 1964, 1974) has proven useful to a diversity of persons, including the physicist Hans Reichenbach in dissolving certain apparent anomalies in quantum mechanics, the electrical engineer Lofti Zadeh in developing fuzzy set theory, and the mathematician Iván Guzmán de Rojas in writing algorithms for the computer-aided analysis and modeling of Aymará grammar. David E. Cooper, a British philosopher, also makes use of Lukasiewicz's three-valued logic, and argues that “Primitive magico-religious thought incorporates an alternative logic to our ‘standard’ one within the terms of which the apparent inconsistencies [noted by anthropologists in certain famous statements made by non-Western peoples] are not inconsistencies at all” (1975:241). In a Reichenbachian variant of three-valued logic, Lukasiewicz's third value, “possible,” is conceptualized preeminantly as indeterminate. Propositions that are indeterminate cannot be either verified or falsified logically, although we may know or believe them to be true or false on other grounds. Thus, for instance, in accordance with the quantum principle of complementarity, some paired propositions complement one another because determining the truth of one renders it impossible in principle to determine the truth of the other. Where verification and falsification are ruled out in principle, a proposition must be assigned some value other than true or false. Cooper holds that when applying three-valued logic in an effort to dissolve anomalies in magico-religious thought, two conditions must be satisfied: First, it must be demonstrated that if “primitive” thought does incorporate this alternative logic, the anomalies do indeed disappear. Second, it must be shown that “primitive” thought actually does incorporate it. If we fulfil the first but not the second condition, we only show that certain troublesome propositions can be interpreted by us in such a way as to render them less troublesome, but we do not save the authors of those propositions from the charge of being inconsistent (and thus, perhaps, irrational). Cooper then goes on to argue in favour of this claim: The magico-religious thought of a people is a highly theoretic explanatory system, within which propositions occur that, while meaningful in terms of the system, are not capable of any verification or falsification within it. Such propositions are not counted by the people in question as being either true or false, but as having a third truth-value. The anomalies arise because the people explicitly reject the consequences of propositions they appear to accept. However, in every such case, we find that at least one of the propositions is counted by the natives, in virtue of its untestability, to be neither true nor false -- hence, despite appearances, the people do not regard as true a number of inconsistent propositions. (Cooper 1975: 244) Cooper's thesis elicited a small number of published responses, for the most part negative. His major examples of non-Western propositions are those that others have also given opinions about, thus lending support to a complaint voiced by Michael Kenny: The recent contributions to the rationality discussion have a pronounced lack of empirical referent, and depend rather heavily on the question of the logic of Azande witchcraft and on Nuer propositions that “twins are birds.” (Kenny 1976: 116) Cooper's interesting thesis, I think, deserves evaluation in fieldwork specifically intended to explore it. Insofar as I am aware, such evaluation has not been undertaken
Dan Sperber 3 August 2010 (00:26)
It is gratifying to get such an interesting discussion, one asking for elaboration of ideas that I have been more commonly called just to make clearer and to defend. Moreover, György Gergely, together with Gergely Csibra and their collaborators, have developed, and provided experimental evidence for important hypotheses about the role of opacity in cultural transmission that converge, I believe, with my own proposal regarding reflective beliefs of semi-propositional content. Moving beyond restating the basics will involve, I am afraid, some technicalities. First, a reminder: The distinction between [i]intuitive[/i] and [i]reflective[/i] beliefs is one between representational format in the mind: intuitive beliefs are mentally represented as plain facts, whereas reflective beliefs are mentally meta-represented as true representations. The distinction between [i]propositional[/i] and [i]semi-propositional[/i] beliefs has to do with their inferential potential: the more directconsequences of propositional belief are easily inferable to the believer and hence the belief can be freely used as a premise in ordinary inferences whereas some of the more direct consequences of a semi-propositional belief are unknown to the believer and therefore using it as a premise in ordinary inferences is likely to cause the inference to stall. Hence the hypothesis that, whereas propositional beliefs may be intuitive or reflective, semi-propositional beliefs can only be reflective: you can only hold them because of some higher order belief about their content, and you cannot use them freely in intuitive inferences. So – György is right –, whether a belief is intuitive or reflective and whether it is propositional or semi-propositional are cognitive-processing distinctions that need not be accessible to introspection. It cannot be decided on the basis of “how it feels”: The phenomenology of beliefs is too vague for that. (Which is not to deny that there is some interesting phenomenology to study in the area.) Now to answer György’s questions: Do, György asks, reflective beliefs imply or require cognitive access to the memory of the source authority? Note first that reflective beliefs need not be based on authority. They may be based on an argument, as when we believe a theorem because we know and understand the proof (and it does not matter whether we discovered the proof on our own or, as is much more common, it was provided by someone else: once we understand the proof, the proof, however we came by it, validates the belief). So György’s question is strictly speaking not about all reflective beliefs, but about reflective beliefs that we accept and go on accepting because of the authority of the source. As he surmises, no, we do not need to remember the source, it is enough that we go on believing that the source, whoever it was, was authoritative. In many case, we hold a reflective belief on the ground that it is widely held by others whom we trust or with whom we identify. We believe it because our teachers/priests/elders do, or because everybody around us does. György suggests that this might involve a different representational format indicating that justification is deferred to some authority (an idea remisniscent of Recanati’s ‘deferential operator’ see for instance here). In fact, deference is ubiquitous not only in belief justification but also in interpretation (and Recanati was mostly referring to the latter kind of cases). Note that if reflective beliefs are, as I claim, metarepresented by being embedded in a validating context, the validating context may consist in an appeal to ‘common knowledge,’ ‘the wisdom of the elders,’ ‘it is said that,’ and so on, that is, do the job for which György suggests there might be a dedicated format. How do we decide between the two proposals? Consider a similar case: Do we have two distinct representational formats in our mind to represent beliefs and to represent desires, or are desires just beliefs that a certain state of affairs would be good? The answer is not a priori obvious. Parsimony would weigh for a single format. However, in the case of beliefs and desires, arguments for assuming that they come in two formats can be drawn from the idea that simpler organisms may have desires without having beliefs (so, parsimony may be lost anyhow), from the fact that I can think that some state of affairs would be good but not desire it (so analyzing desires as beliefs with a specific kind of content is not easily done), and so on. All these arguments are sensible (and I find them, taken together, convincing) but they are not undisputable. In the case of reflective beliefs accepted because of the authority of source without however any clear and lasting identification of that source, should we see them as involving a distinct representational format? Here parsimony seems to me to weigh against this extra format: I don’t see that the general metarepresentational format of reflective beliefs falls short of providing a good means for handling these cases. This however is not a knock-down argument. I guess, moreover, that what motivates György is the hunch that opaque contents are, from infancy, processed in a distinct way. Note however that not all beliefs accepted because of the authority of the source are opaque or ‘semi-propositional’. I may understand and believe a theorem, say Fermat’s last theorem, without understanding the proof, hence believe it because of the authority of the source and I may do so do so without remembering what the source of my belief was, hence hold a reflective belief of truly propositional, non-opaque content. I imagine that in such a case, György would not be tempted to suggest that I hold the theorem in a special format. The real issue is not so much ill-remembered authority as poorly understood content. The general issue that György is raising is whether there are categorically distinct attitudes in approaching different subtypes of opaque beliefs based on authority. He illustrates the hypothetical contrast with the cases of: (a) “there are negative numbers” (b) “the word ‘stone’ refers to stones” both beliefs being accepted because of the authority of the source, but, he suggest, the first one typically with an identified source, for instance the teacher, and the second one with just the authority of general knowledge. I would argue deference to general knowledge might be enough to accept belief (a) about negative numbers, and that (b) is not so much a belief as a skill. Learning the word ‘stone’ is learning to understand it as referring to stones and learning to use it to refer to stones. It is not acquiring a metalinguistic belief about the word 'stone'. It takes a philosopher or a linguist to believe (b) (or to question it). Regarding other skills involved in the use of artifact such as spoons and knives, I am not sure how much and how early they involve actual beliefs about function (and when I say I am not sure, this is not a manner of speaking: I genuinely don’t have clear views on the matter). Are there, asks György, two types of opacity or is there just one? Does the child presented with obscure religious statements treat them from the start as different from obscure technical statement, as if she knew that the former could never become fully propositional whereas the latter should? I am willing to entertain the idea, but I do find it puzzling. How would the child know which is which? And isn’t this to rigidify the distinction more than I would and much more than Maurice would? After all, some statements, e.g. scientific ones, become fully propositional for some people while remaining semi-propositional for others. Do young children ‘know’ whether they will develop scientific interest and competencies and what is likely therefore to become or not to become propositional for them individually? More importantly, from an epidemiological point of view, I would argue that, to the developing mind, statements that are irredeemably semi-propositional – ‘religious’ statements in particular – have no signature difference from statements that turn out to be fully interpretable. This fact plays a crucial role in the capacity of religious beliefs to invade human populations: they meet the input condition of the cognitive receptors for opaque information that is likely to become transparent and highly relevant. By the time – adolescence ? – it turns out that they will remain mysterious, they are well rooted and, anyhow, have acquired other forms of cognitive and social relevance. All this is quite difficult, and thanks for the occasion to try and plod forward.