Homeostatic property cluster kinds
A general account of what a natural kind is in the special sciences, due to Richard Boyd (1991), imported into emotion theory by Scarantino (2018). It is on the wiki because it is the hinge of the “natural kinds” dispute that runs through basic-emotions — and because both sides of that dispute have been arguing past it.
The contrast
Essentialist kinds “must possess definitional essences that define them in terms of necessary and sufficient, intrinsic, unchanging, ahistorical properties” (Boyd, 146). Water is H₂O; every sample, always, regardless of history.
HPC kinds are defined instead by:
- clusters of properties that co-occur — in an important number of cases, not all;
- properties that are relational, changing, and historical rather than intrinsic and fixed;
- co-occurrence produced by a multiplicity of imperfect homeostatic mechanisms, not by an essence;
- and — the constraint that keeps this from being vacuous — the cluster must support theoretically interesting generalizations on account of that co-occurrence.
Boyd’s argument, which Scarantino leans on rather than reproduces: these are the only natural kinds the special sciences are likely to find. It is commonplace there to have multiple combinations of properties sufficient for membership and no properties necessary for it. The paradigm case is biological species — no genetic or morphological essence, historically constituted, internally variable, and unquestionably a joint at which nature is carved.
Why an emotion theorist needs this
The empirical situation Scarantino inherits, and concedes:
- Instances of anger, fear, happiness and so on are “massively variable with respect to facial movements, autonomic changes, and preset and learned actions.”
- Therefore folk emotion categories “do not designate natural kinds: their extensions are too heterogenous for any scientifically interesting generalizations to be true of all of their members.”
On an essentialist conception of kinds, that record is fatal to basic emotion theory — which is exactly how Barrett’s “are emotions natural kinds?” argument is usually taken, and how Lindquist et al. (2012)‘s failure to find consistency-and-specificity is usually read.
On an HPC conception, the record is what the theory predicts. Scarantino offers basic emotions as HPC kinds that
lack essences and fixed boundaries, show massive internal variability due to the need of adapting responses to changing circumstances, lack one-to-one mappings with any folk psychological categories, but nevertheless carve nature at its joints in the sense that they are instantiated by clusters of theoretically important properties that co-occur on account of underlying causal mechanisms (affect programs).
The variability has a reason on this account, and the reason is adaptive: responses must fit circumstances, so a program that produced identical outputs would be worse at its job. This dovetails with the input-output open revision — flexibility of output is the mechanism that generates the variability that HPC kinds tolerate.
What it does to the debate
This is the move that converts the constructionist’s strongest evidence into the nativist’s expectation, and the wiki should record how much it costs and what it does not buy.
What it concedes (more than Friedman concedes): folk categories are not natural kinds; the bodily signatures are not there and were never coming; emotions have no essences.
What it keeps: evolved, pan-cultural, causally potent affect programs with homologs in other species — the “hard core” of BET in Lakatos’ sense, with the essentialism relegated to the “protective belt” of auxiliary assumptions that a research programme may modify without abandoning.
What it does not settle: whether basic fear actually is an HPC kind. That requires meeting the fruitfulness constraint — showing the cluster supports scientifically interesting generalizations — and Scarantino asserts this and cites his earlier work ([3]) rather than demonstrating it here. Without that step the framework is unfalsifiable-looking, and this is flagged as a limitation on the study page.
The two ways to criticize an explication
Worth recording alongside, because it is the general apparatus (Carnap 1950) within which Scarantino operates and it applies to any category the wiki treats as constructed-for-purpose — including interoception’s own seven-construct taxonomy:
- Similarity problem — the explicated category is not similar enough to the folk category it transforms. (Is basic fear recognizably fear?)
- Fruitfulness problem — the explicated category supports no scientifically interesting generalizations. (Does anything true hold of all basic fear?)
Crucially, on Scarantino’s reading, the variability of folk categories speaks to neither: “what matters is whether the instances that do meet the explicative definition of basic emotion are similar enough to emotions as folk psychologically understood and, most importantly, whether they can be embedded in generalizations that are scientifically interesting.”
This is what licenses his reply to critics: “A critique of a theory of basic fear based on data about fear writ large is like a critique of a theory of short-term memory based on empirical data about memory writ large.” The memory precedent is the model throughout — folk “memory” turned out not to be theoretically homogeneous (Schacter), which did not abolish memory science; it produced short-term and long-term memory, investigated independently.
An open worry for the wiki
The escape hatch is wide. A kind with no necessary properties, no fixed boundaries, and expected massive variability is difficult to disconfirm, and the only thing standing between HPC-BET and unfalsifiability is the fruitfulness requirement — which is the one part not discharged in the source. See what-should-emotion-terms-refer-to for the same structure appearing elsewhere in the paper: a strong general principle, with the load-bearing demonstration deferred to prior work.