Cimpian & Salomon (C&S) posit an inherence heuristic to explicate how humans explain patterns of co-occurrence. Their proposal is that the heuristic requires explanations to be couched in terms of the inherent features of the objects that figure in the co-occurrence pattern. So, for example, that lions co-occur with lion properties is to be explained in terms of the inherent features of lions and their properties. C&S argue that explanations of this sort are precursors to psychological essentialism, the view that a category's properties are determined by its essence, and that people's beliefs and behaviours are consistent with this.
However, C&S's proposal appears to come unstuck on precisely the claim that inherence is an heuristic. Heuristics are fast and frugal cognitive mechanisms (Gigerenzer et al., Reference Gigerenzer and Todd1999) that deliver simple and shallow outputs – outputs that are right most of the time (cf. bounded rationality [Simon Reference Simon1957]). These properties provide a plausible basis for heuristics' adaptive value: Heuristics have evolved because they are simple and fast, and because they generally deliver the right answers. In effect, heuristics succeed because they transform problems that are difficult to solve and computationally complex into ones that are easier to solve and computationally simple (Kahneman & Frederick Reference Kahneman, Frederick, Gilovich, Griffin and Kahneman2002; Tversky & Kahneman Reference Tversky and Kahneman1973).
However, a careful unpacking of C&S's account indicates that the inherence heuristic requires cognitive outputs to be neither quick, nor shallow, nor adaptive.
Take the co-occurrence of lions and lion properties. According to the inherence heuristic, an explanation of this co-occurrence will involve finding a story that explains why the inherent features of lions and the inherent features of lion properties co-occur. Of course, such an explanation is not obvious. Indeed, C&S acknowledge that the heuristic may deliver the outcome that the to-be-explained co-occurrence derives “in some to-be-determined fashion” from a “to-be-determined combination of inherent features” (sect. 4.3, para. 1) (consistent with the argument that categories may be associated with an essence “placeholder” [sect. 4.3, para. 6]).
Given that heuristics are thought to operate quickly, deliver shallow outputs, and offer adaptive value, we need to ask how the inherence heuristic fares on these criteria.
For lions and lion properties, the search for a co-occurrence explanation will ultimately terminate either in an explanation that is approximately true (e.g., lion DNA), or one that is false, or one that is a placeholder. The true explanation may arguably be adaptive, but would be neither quick nor shallow. The false explanation could be quick and shallow but, being false, is hardly likely to be adaptive. The placeholder explanation can be neither quick nor shallow and is of questionable adaptive value. Thus, either way you look at it, the inherence heuristic delivers outputs that imply it is not an heuristic after all.
Indeed, the inherence heuristic proposes people do very considerable cognitive work of very little value, inconsistent with the idea that heuristics turn computationally complex problems into simpler ones. Take the very regular co-occurrence between the sun and the location of its rising: From a fixed location, the sun always seems to rise in the same place – on the East coast the sun always rises over the sea. According to the inherence heuristic, people seek an explanation for this in terms of the inherent properties of the sun and the sea (or that part of the sea where the sun breaches the horizon). But such a search would be fruitless. The sun rises where it does because of its relation to the Earth, not because of the inherent properties of either. The inherence heuristic would ultimately generate either a false or an inchoate explanation. Crucially, it could not be quick and shallow and adaptive. Take another example where relational structure is central: the very regular co-occurrence between liquids and their direction of flow – liquids flow downhill. The inherence heuristic implies people strive to explain this in terms of the inherent properties of liquids and the ground on which they flow. The heuristic entirely misses the relation (gravitation) essential to a successful explanation, and thereby must predict that people ultimately generate either false or inchoate explanations.
These kinds of example could be given computationally simpler explanations by an heuristic mechanism. A view that C&S briefly acknowledge, but ultimately eschew, is that successful explanations might advert to “brute statistical facts” (sect. 1, para. 4). Why do lions and lion properties co-occur? Why does the sun rise where it does? Why do liquids flow downhill? Because that is the way things are; because these are just brute facts. Unlike the inherence heuristic, a brute-fact heuristic would appear genuinely to be quick, shallow and adaptive. Whether brute facts exist is a matter for philosophical analysis. The point is that, whether or not they do, there is nothing to stop an heuristic device presupposing their existence in explaining co-occurrence patterns.
Strangely, C&S's proposal is not so inconsistent with brute-fact explanations. It relies on the claim that stable, inherent properties of objects are laid down in semantic memory before co-occurrence explanations are sought. It is as if the inherence heuristic actually relies on a prior stage where properties of a to-be-learned category are processed simply as if they were brute facts, no explanation required, no questions asked. But if inherence can tolerate brute facts in laying down semantic memories, it is puzzling indeed that C&S reject it in explaining co-occurrence.
Human history is replete with examples where explanations for earlier categorisations have been sought, and categorisations challenged or overturned (e.g., Eco Reference Eco1999). In this, the inherent features of categories and their properties figure strongly, but people also go beyond inherent features and look to relations, too. Likewise, searching for an explanation of categorisation in terms of essential properties is likely to be neither quick nor computationally simple. The point is that these societal processes are complex. No doubt there must be intra-personal analogues, but the lesson of human history is that, to do the job, explanations will be computationally complex and deliver deep outputs. A quick and simple heuristic is just not up to the job – it is at once entirely too fast and entirely too easy.
You have
Access
Cimpian & Salomon (C&S) posit an inherence heuristic to explicate how humans explain patterns of co-occurrence. Their proposal is that the heuristic requires explanations to be couched in terms of the inherent features of the objects that figure in the co-occurrence pattern. So, for example, that lions co-occur with lion properties is to be explained in terms of the inherent features of lions and their properties. C&S argue that explanations of this sort are precursors to psychological essentialism, the view that a category's properties are determined by its essence, and that people's beliefs and behaviours are consistent with this.
However, C&S's proposal appears to come unstuck on precisely the claim that inherence is an heuristic. Heuristics are fast and frugal cognitive mechanisms (Gigerenzer et al., Reference Gigerenzer and Todd1999) that deliver simple and shallow outputs – outputs that are right most of the time (cf. bounded rationality [Simon Reference Simon1957]). These properties provide a plausible basis for heuristics' adaptive value: Heuristics have evolved because they are simple and fast, and because they generally deliver the right answers. In effect, heuristics succeed because they transform problems that are difficult to solve and computationally complex into ones that are easier to solve and computationally simple (Kahneman & Frederick Reference Kahneman, Frederick, Gilovich, Griffin and Kahneman2002; Tversky & Kahneman Reference Tversky and Kahneman1973).
However, a careful unpacking of C&S's account indicates that the inherence heuristic requires cognitive outputs to be neither quick, nor shallow, nor adaptive.
Take the co-occurrence of lions and lion properties. According to the inherence heuristic, an explanation of this co-occurrence will involve finding a story that explains why the inherent features of lions and the inherent features of lion properties co-occur. Of course, such an explanation is not obvious. Indeed, C&S acknowledge that the heuristic may deliver the outcome that the to-be-explained co-occurrence derives “in some to-be-determined fashion” from a “to-be-determined combination of inherent features” (sect. 4.3, para. 1) (consistent with the argument that categories may be associated with an essence “placeholder” [sect. 4.3, para. 6]).
Given that heuristics are thought to operate quickly, deliver shallow outputs, and offer adaptive value, we need to ask how the inherence heuristic fares on these criteria.
For lions and lion properties, the search for a co-occurrence explanation will ultimately terminate either in an explanation that is approximately true (e.g., lion DNA), or one that is false, or one that is a placeholder. The true explanation may arguably be adaptive, but would be neither quick nor shallow. The false explanation could be quick and shallow but, being false, is hardly likely to be adaptive. The placeholder explanation can be neither quick nor shallow and is of questionable adaptive value. Thus, either way you look at it, the inherence heuristic delivers outputs that imply it is not an heuristic after all.
Indeed, the inherence heuristic proposes people do very considerable cognitive work of very little value, inconsistent with the idea that heuristics turn computationally complex problems into simpler ones. Take the very regular co-occurrence between the sun and the location of its rising: From a fixed location, the sun always seems to rise in the same place – on the East coast the sun always rises over the sea. According to the inherence heuristic, people seek an explanation for this in terms of the inherent properties of the sun and the sea (or that part of the sea where the sun breaches the horizon). But such a search would be fruitless. The sun rises where it does because of its relation to the Earth, not because of the inherent properties of either. The inherence heuristic would ultimately generate either a false or an inchoate explanation. Crucially, it could not be quick and shallow and adaptive. Take another example where relational structure is central: the very regular co-occurrence between liquids and their direction of flow – liquids flow downhill. The inherence heuristic implies people strive to explain this in terms of the inherent properties of liquids and the ground on which they flow. The heuristic entirely misses the relation (gravitation) essential to a successful explanation, and thereby must predict that people ultimately generate either false or inchoate explanations.
These kinds of example could be given computationally simpler explanations by an heuristic mechanism. A view that C&S briefly acknowledge, but ultimately eschew, is that successful explanations might advert to “brute statistical facts” (sect. 1, para. 4). Why do lions and lion properties co-occur? Why does the sun rise where it does? Why do liquids flow downhill? Because that is the way things are; because these are just brute facts. Unlike the inherence heuristic, a brute-fact heuristic would appear genuinely to be quick, shallow and adaptive. Whether brute facts exist is a matter for philosophical analysis. The point is that, whether or not they do, there is nothing to stop an heuristic device presupposing their existence in explaining co-occurrence patterns.
Strangely, C&S's proposal is not so inconsistent with brute-fact explanations. It relies on the claim that stable, inherent properties of objects are laid down in semantic memory before co-occurrence explanations are sought. It is as if the inherence heuristic actually relies on a prior stage where properties of a to-be-learned category are processed simply as if they were brute facts, no explanation required, no questions asked. But if inherence can tolerate brute facts in laying down semantic memories, it is puzzling indeed that C&S reject it in explaining co-occurrence.
Human history is replete with examples where explanations for earlier categorisations have been sought, and categorisations challenged or overturned (e.g., Eco Reference Eco1999). In this, the inherent features of categories and their properties figure strongly, but people also go beyond inherent features and look to relations, too. Likewise, searching for an explanation of categorisation in terms of essential properties is likely to be neither quick nor computationally simple. The point is that these societal processes are complex. No doubt there must be intra-personal analogues, but the lesson of human history is that, to do the job, explanations will be computationally complex and deliver deep outputs. A quick and simple heuristic is just not up to the job – it is at once entirely too fast and entirely too easy.