The inherence heuristic captures a deep characteristic of commonsense thought. As currently formulated, however, it is limited by an extremely cursory characterization of inherent features and patterns. A more robust characterization of the patterns that are grist for the inherence heuristic, as well as how inherent features are represented, would greatly improve the heuristic's ability to predict the types of intuitions that may be generated when explaining a specific pattern. Recent work on the representation of generic knowledge can help provide this more robust characterization.
It is noteworthy that the patterns discussed by Cimpian & Salomon (C&S) involve kinds of things such as boys, girls, and orange juice. This is not an accident. That all coins in my pocket on a given day are copper does not constitute a pattern that can be explained via the inherence heuristic. Such a pattern involves an accidental generalization and thus cannot be explained, much less explained via inherent properties of the coins in my pocket or my pocket. Thus, a minimal condition for the operation of the inherence heuristic is that the pattern be one that involves a nonaccidental generalization and thus is extendable to indefinitely many new instances of the relevant sort (Goodman Reference Goodman1955/1983).
Prasada et al. (Reference Prasada, Khemlani, Leslie and Glucksberg2013) provide evidence that our conceptual systems distinguish at least three types of nonaccidental connections between kinds (which contain indefinitely many instances) and properties. Kinds may be characterized by properties that have a principled, causal, or statistical connection to the kind. Each connection type grounds distinct linguistic and nonlinguistic phenomena and reflects a distinct perspective from which we can think about kinds of things. Properties that have a principled or a casual connection to a kind may plausibly be considered inherent properties of the kind when thinking about kinds from a formal or material perspective, respectively, and ground distinct types of inherent thinking.
Properties that have a principled connection to a kind are properties that instances of a kind are understood to have by virtue of their being the kinds of things they are (k-properties). K-properties are properties (1) whose presence in instances of a kind support formal explanations – explanations by reference to the kind of thing something is (e.g., Fido has four legs because he is a dog); (2) for which we have normative expectations such that instances of the kind that lack them are judged to be defective or incomplete; and (3) are generally expected to be present in instances of the kind (Prasada & Dillingham Reference Prasada and Dillingham2006; Reference Prasada and Dillingham2009). K-properties are understood to be an aspect of being the kind of thing in question and thus are represented via a formal part–whole relation between the kind and property (Prasada & Dillingham Reference Prasada and Dillingham2009).
Attending to principled connections brings forth the formal dimension of our commonsense conceptions, and we notice formal explanatory relations, as well as the basis for certain normative and statistical expectations. In so doing, k-properties can ground key aspects of inherent thinking identified by C&S. Specifically, the intuition that the pattern is inevitable reflects our expectation that k-properties will generally be present in instances of the kind. K-properties also ground the normative dimension of much inherent thinking – dogs are supposed to have four legs and the ones that don't have something wrong with them. Note that this is stronger than simply the intuition that something is good or beneficial. We think that it is beneficial for dogs to wear collars, but we don't think there is anything wrong with dogs that don't (Bublitz & Prasada Reference Bublitz and Prasada2013).
Attending to casual connections between kinds and properties brings forth the material dimension of our commonsense conceptions by focusing on the material constitution of the instances of kinds and thus their casual dispositions to behave in one or another manner in appropriate circumstances. Properties that merely have a casual connection to a kind do not ground an expectation that they are generally present in instances of the kind, nor do they ground normative expectations about their presence in instances of the kind (Prasada et al. Reference Prasada, Khemlani, Leslie and Glucksberg2013). As such, though properties that have a casual connection to a kind may be considered inherent properties of the kind and can ground causal explanations of patterns, the intuitions of inevitability and normativity associated with much inherent thinking cannot derive from our understanding of these properties having a casual connection to the kind.
To account for the range of characteristics associated with inherent thinking discussed in the article by C&S, the “storytelling” mechanism must crucially have access to information as to whether a property is considered an inherent property because it is represented as an aspect of being that kind of thing and thus has a principled connection to the kind or if the property is considered an inherent property of the kind because it is casually connected to its material constitution. The type of connection represented between the kind and the property is needed to determine whether the outcome of the storytelling process will include intuitions of inevitability and a normative dimension and whether it will ground causal explanations. Many properties will have both principled and causal connections to a kind (e.g., the four-leggedness of dogs)
C&S are correct that the work on k-properties does not provide an alternative to the inherence heuristic; however, if the present argument is correct, it is a necessary complement to the inherence heuristic. Incorporating the insights from the work on the representation of generic knowledge into the inherence heuristic promises to be a fruitful avenue for future research. Detailed work on the representation of different kind of “habituals” (e.g., John takes the train to work vs. John prefers blonds) will also likely help inform how the inherence heuristic functions when reasoning about instances rather than patterns. Finally, progress on specifying the scope of the inherence heuristic can likely be made by detailed investigations of the formal characteristics that distinguish kind representations from representations of other types of multiplicities (Prasada et al. Reference Prasada, Hennefield and Otap2012).
The inherence heuristic captures a deep characteristic of commonsense thought. As currently formulated, however, it is limited by an extremely cursory characterization of inherent features and patterns. A more robust characterization of the patterns that are grist for the inherence heuristic, as well as how inherent features are represented, would greatly improve the heuristic's ability to predict the types of intuitions that may be generated when explaining a specific pattern. Recent work on the representation of generic knowledge can help provide this more robust characterization.
It is noteworthy that the patterns discussed by Cimpian & Salomon (C&S) involve kinds of things such as boys, girls, and orange juice. This is not an accident. That all coins in my pocket on a given day are copper does not constitute a pattern that can be explained via the inherence heuristic. Such a pattern involves an accidental generalization and thus cannot be explained, much less explained via inherent properties of the coins in my pocket or my pocket. Thus, a minimal condition for the operation of the inherence heuristic is that the pattern be one that involves a nonaccidental generalization and thus is extendable to indefinitely many new instances of the relevant sort (Goodman Reference Goodman1955/1983).
Prasada et al. (Reference Prasada, Khemlani, Leslie and Glucksberg2013) provide evidence that our conceptual systems distinguish at least three types of nonaccidental connections between kinds (which contain indefinitely many instances) and properties. Kinds may be characterized by properties that have a principled, causal, or statistical connection to the kind. Each connection type grounds distinct linguistic and nonlinguistic phenomena and reflects a distinct perspective from which we can think about kinds of things. Properties that have a principled or a casual connection to a kind may plausibly be considered inherent properties of the kind when thinking about kinds from a formal or material perspective, respectively, and ground distinct types of inherent thinking.
Properties that have a principled connection to a kind are properties that instances of a kind are understood to have by virtue of their being the kinds of things they are (k-properties). K-properties are properties (1) whose presence in instances of a kind support formal explanations – explanations by reference to the kind of thing something is (e.g., Fido has four legs because he is a dog); (2) for which we have normative expectations such that instances of the kind that lack them are judged to be defective or incomplete; and (3) are generally expected to be present in instances of the kind (Prasada & Dillingham Reference Prasada and Dillingham2006; Reference Prasada and Dillingham2009). K-properties are understood to be an aspect of being the kind of thing in question and thus are represented via a formal part–whole relation between the kind and property (Prasada & Dillingham Reference Prasada and Dillingham2009).
Attending to principled connections brings forth the formal dimension of our commonsense conceptions, and we notice formal explanatory relations, as well as the basis for certain normative and statistical expectations. In so doing, k-properties can ground key aspects of inherent thinking identified by C&S. Specifically, the intuition that the pattern is inevitable reflects our expectation that k-properties will generally be present in instances of the kind. K-properties also ground the normative dimension of much inherent thinking – dogs are supposed to have four legs and the ones that don't have something wrong with them. Note that this is stronger than simply the intuition that something is good or beneficial. We think that it is beneficial for dogs to wear collars, but we don't think there is anything wrong with dogs that don't (Bublitz & Prasada Reference Bublitz and Prasada2013).
Attending to casual connections between kinds and properties brings forth the material dimension of our commonsense conceptions by focusing on the material constitution of the instances of kinds and thus their casual dispositions to behave in one or another manner in appropriate circumstances. Properties that merely have a casual connection to a kind do not ground an expectation that they are generally present in instances of the kind, nor do they ground normative expectations about their presence in instances of the kind (Prasada et al. Reference Prasada, Khemlani, Leslie and Glucksberg2013). As such, though properties that have a casual connection to a kind may be considered inherent properties of the kind and can ground causal explanations of patterns, the intuitions of inevitability and normativity associated with much inherent thinking cannot derive from our understanding of these properties having a casual connection to the kind.
To account for the range of characteristics associated with inherent thinking discussed in the article by C&S, the “storytelling” mechanism must crucially have access to information as to whether a property is considered an inherent property because it is represented as an aspect of being that kind of thing and thus has a principled connection to the kind or if the property is considered an inherent property of the kind because it is casually connected to its material constitution. The type of connection represented between the kind and the property is needed to determine whether the outcome of the storytelling process will include intuitions of inevitability and a normative dimension and whether it will ground causal explanations. Many properties will have both principled and causal connections to a kind (e.g., the four-leggedness of dogs)
C&S are correct that the work on k-properties does not provide an alternative to the inherence heuristic; however, if the present argument is correct, it is a necessary complement to the inherence heuristic. Incorporating the insights from the work on the representation of generic knowledge into the inherence heuristic promises to be a fruitful avenue for future research. Detailed work on the representation of different kind of “habituals” (e.g., John takes the train to work vs. John prefers blonds) will also likely help inform how the inherence heuristic functions when reasoning about instances rather than patterns. Finally, progress on specifying the scope of the inherence heuristic can likely be made by detailed investigations of the formal characteristics that distinguish kind representations from representations of other types of multiplicities (Prasada et al. Reference Prasada, Hennefield and Otap2012).