3650 words

Introduction

For more than 100 years—from Galton and Spearman to Burt, Jensen, Rushton, Lynn and today’s polygenic score enthusiasts—hereditarian thinkers have argued that general intelligence is a unitary, highly heritable biological trait and that observed individual and group level differences in IQ and it’s underlying “g” factor primarily reflect genetic causation. The Bell Curve brought such thinking into the mainstream from obscure psychology journals, and today hereditarian behavioral geneticists claim that 10 to 20 percent of the variance in education and cognitive performance has been explained by GWA studies (see Richardson, 2017). The consensus is that intelligence within and between populations is largely genetic in nature.

While hereditarianism is empirically contested and morally wrong, the biggest kill-shot is that it is conceptually impossible, and one can use many a priori arguments from philosophy of mind to show this. Donald Davidson’s argument against the possibility of psychophysical laws, Kripke’s reading of Wittgenstein, and Nagel’s argument from indexicality can be used to show that hereditarianism is a category error. Ken Richardson’s systems theory can then be used to show that g is an artifact of dynamic systems (along with test construction), and Vygotsky’s cultural-historical psychology shows that higher mental functions (which hereditarians try to explain biologically) originate as socially scaffolded, inter-mental processes mediated by cultural tools and interactions with more knowledgeable others, not individual genetic endowment.

Thus, these metaphysical, normative, systemic, developmental and phenomenological refutations show that hereditarianism is based on a category mistake. Ultimately, what hereditarianism lacks is a coherent object to measure—since psychological traits aren’t measurable at all. I will show here how hereditarianism can be refuted with nothing but a priori logic, and then show what really causes differences in test scores within and between groups. Kripke’s Wittgenstein and the argument against the possibility of psychophysical laws, along with a Kim-Kripke normativity argument against hereditarianism show that hereditarianism just isn’t a logically tenable position. So if it’s not logically tenable, then the only way to explain gaps in IQ is an environmental one.

I will begin with showing that no strict psychophysical laws can link genes/brain states to mental kinds, then demonstrating that even the weaker functional-reduction route collapses at the very first step because no causal-role definition of intentionality (intelligence) is possible. After that I will add the general rule following considerations from Kripke’s Wittgenstein and then add it to my definition of intelligence, showing that rule-following is irreducibly normative and cannot be fixed by any internal state and that no causal-functional definition is possible. Then I will show that the empirical target of hereditarianism—the g factor—is nothing more than a statistical artifact of historically contingent, culturally-situated rule systems and not a biological substrate. These rule systems do not originate internally, but they develop as inter-mental relations mediated by cultural tools. Each of these arguments dispenses with attempted hereditarian escapes—the very notion of a genetically constituted, rank-orderable general intelligence is logically impossible.

We don’t need “better data”—I will demonstrate that the target of hereditarian research does not and cannot exist as a natural, measurable, genetically-distributed trait. IQ scores are not measurements of a psychological magnitude (Berka, 1983; Nash, 1990); no psychophysical laws exist that can bridge genes to normative mental kinds (Davidson, 1979), and the so-called positive manifold is nothing more than a cultural artifact due to test construction (Richardson, 2017). Thus, what explains IQ variance is exposure to the culture in which the right rules are used regarding the IQ test.

Psychophysical laws don’t exist

Hereditarianism implicitly assumes a psychophysical law like “G -> P.” Psychophysical laws are universal, necessary mappings between physical states and mental states. To reduce the mental to the physical, you need lawlike correlations—whenever physical state P occurs, mental state M occurs. These laws must be necessary, not contingent. They must bridge the explanatory gap from the third-personal to the first-personal. We have correlations, but correlations don’t entail identity. If correlations don’t entail identity, then the correlations aren’t evidence of any kind is psychophysical law. So if there are no psychophysical laws, there is no reduction and there is no explanation of the mental.

Hereditarianism assumes type-type psychophysical reduction. Type-type identity posits that all instances of a mental type correspond to all instances of a physical type. But hereditarians need bridge laws—they imply universal mappings allowing reduction of the mental to the measurable physical. But since mental kinds are anomalous, type-type reduction is impossible.

Hereditarians claim that genes cause g which then cause intelligence. This requires type-type reduction. Intelligence kind = g kind = physical kind. But g isn’t physical—it’s a mathematical construct, the first PC. Only physical kinds can be influenced by genes;nonphysical kinds cannot. Even if g correlates with brain states, correlation isn’t identity. Basically, no psychophysical laws means no reduction and therefore no mental explanation.

If hereditarianism is true, then intelligence is type-reducible to g/genes. If type-reduction holds, then strict psychophysical laws exist. So if hereditarianism is true, then strict psychophysical laws exist. But no psychophysical laws exist, due to multiple realizablilty and Davidson’s considerations. So hereditarianism is false.

We know that the same mental kind can be realized in different physical kinds, meaning that no physical kind correlates one-to-one necessarily with a mental kind. Even if we generously weaken the demand from strict identity to functional laws, hereditarian reduction still fails (see below).

The Kim-Kripke normativity argument

Even the only plausible route to mind-body reduction that most physicalists still defend collapses a priori for intentional/cognitive states because no causal-functional definition can ever capture the normativity of meaning and rule following (Heikenhimo, 2008). Identity claims like water = h2O only work because the functional profile is already reducible. Since the functional profile of intentional intelligence is not reducible, there is no explanatory bridge from neural states to the normativity of thought. So identity claims fail—this just strengthens Davidson’s conclusions. Therefore, every reductionist strategy that could possibly license the move from “genetic variance -> variation in intelligence” is blocked a priori.

(1) If hereditarianism is true, then general intelligence as a real cognitive capacity must be reducible to the physical domain (genes, neural states, etc).

(2) The only remaining respectable route to mind-body reduction of cognitive/intentional processes is Kim’s three-step functional-reduction model.

(C1) So if hereditarianism is true, then general intelligence must he reducible to Kim’s three-step functional-reduction model.

(3) Kim-style reduction requires—as its indispensable first step—an adequate causal-functional definition of the target property (intelligence, rule-following, grasping meaning, etc) that preserves the established normative meaning of the concept without circularly using mental/intentional vocabulary in the definiens.

(4) Any causal-functional definition of intentional/cognitive states necessarily obliterates the normative distinction between correct and incorrect application (Kripke’s normativity argument applied to mental content).

(C2) Therefore, no adequate causal-functional definition of general intelligence is possible, even in principle.

(5) If no adequate causal-functional definition is possible, then Kim-style functional reduction of general intelligence is impossible.

(C3) So Kim-style functional reduction of general intelligence is impossible.

(C4) So hereditarianism is false.

A hereditarian can resist Kim-Kripke in 4 ways but each fails. (1) They can claim intelligence need not be reducible, but then genes cannot causally affect it, dissolving hereditarianism into mere correlation. (2) They can reject Kim-style reduction in factor of non-reductive or mechanistic physicalism, but these views still require functional roles and collapse under Kim’s causal exclusion argument. (3) They can insist that intelligence has a purely causal-functional definition (processing efficiency or pattern recognition), but such definitions omit the normativity of reasoning and therefore do no capture intelligence at all. (4) They can deny that normativity matters, but removing correctness conditions eliminates psychological content and makes “intelligence” unintelligible, destroying the very trait hereditarianism requires. Thus, all possible routes collapse into contradiction or eliminativism.

The rule-following argument against hereditarianism

Imagine a child who is just learning to add. She adds 68+57=125. We then say that she is correct. Why is 125 correct and 15 incorrect? It isn’t correct because she feels sure, because someone who writes 15 could be just as sure. It isn’t correct because her brain lit up in a certain way, because the neural pattern could also belong to someone following a different rule. It isn’t correct because all of her past answers, because all past uses were finite and are compatible with infinitely many bizzare rules that only diverge now. It isn’t correct because of her genes or any internal biological state, because DNA is just another finite physical fact inside of her body.

There is nothing inside of her head, body or genome that reaches out and touches the difference between correct and incorrect. But the difference is real. So where does it lie? It lives outside of her in the shared community practices. Correctness is a public status, not a private possession. Every single thing that IQ tests reward—series completion, analogies, classification, vocabulary, matrix reasoning—is exactly this kind of going on correctly. So every single point on an IQ test is an act whose rightness is fixed in the space of communal practice. What we call “intelligence” exists only between us—between the community, society and culture in which an individual is raised.

Intelligence is a normative ability. To be intelligent is to go on in the same way, to apply concepts correctly, to get it right when solving new problems, reasoning, understanding analogies, etc. So intelligence = rule-following (grasping and correctly applying abstract patterns).

Rule following is essentially normative—there is a difference between seeming right and being right. Any finite set of past performances is compatible with an infinite set of many rules. No fact about an individual—neither physical nor mental content—uniquely determines the rule they are following. So no internal state fixes the norm. Thus, rule following cannot be constituted by internal/genetic states. No psychophysical law can connect G to correct rule following (intelligence).

Therefore rule-following is set by participation in a social practice. Therefore, normative abilities (intelligence, reasoning, understanding) are socially, not genetically, constituted. So hereditarianism is logically impossible.

At its core, intelligence is the ability to get it right. Getting it right is a social status conferred by participation in communal practices. No amount of genetic or neural causation can confer that status—because no internal state can fix the normative fact. So the very concept of “genetically constituted general intelligence” is incoherent. Therefore, hereditarianism is logically impossible.

(1) H -> G -> P
Hereditarianism -> genes/g -> normative intelligence
(2) P -> R
Normative intelligence -> correct rule-following.
(3) R -> ~G
Rule following cannot be fixed by internal physical/mental states.
So ~(G -> P)
So ~H.

The Berka-Nash measurement objection

This is a little-known critique of psychology and IQ. First put forth in Karel Berka’s 1983 book Measurement: It’s Concepts, Theories, and Problems, and then elaborated on in Roy Nash’s (1990) Intelligence and Realism: A Materialist Critique of IQ.

If hereditarianism is true, then intelligence must be a measurable trait (with additive structure, object, and units) that genes can causally influence via g. If intelligence is measurable, then psychophysical laws must exist to map physical causes to mental kinds. But no such measurability or laws exist. Thus, hereditarianism is false.

None of the main, big-name hereditarians have ever addressed this type of argument. (Although Brand et al, 2003 did attempt to, their critique didn’t work and they didn’t even touch the heart of the matter.) Clearly, the argument shows that hereditarian psychology is weak to such critique. The above argument shows that IQ is quasi-quantification, without an empirical object, no structure, or lawful properties

The argument for g is circular

“Subtests within a battery of intelligence tests are included n the basis of them showing a substantial correlation with the test as a whole, and tests which do not show such correlations are excluded.” (Tyson, Jones, and Elcock, 2011: 67)

g is defined as the common variance of pre-selected subtests that must correlate. Subtests are included only if they correlate. A pattern guaranteed by construction cannot be evidence of a pre-existing biological unity. So g is a tautological artifact, not a natural kind that genes can cause.

Hereditarians need g to be a natural kind trait that genes can act upon. But g is an epiphenomenal artifact due to test construction produced by current covariation of culturally specific cognitive tasks in modern school societies. Artifacts of historically contingent cultural ecologies are not natural kind traits. So g is not a natural kind. So hereditarianism is false.

The category error argument

Intelligence is a first-person indexical act. g is a third-person statistical abstraction. There can be no identity between a phenomenonal act and a statistical abstraction. So g cannot be intelligence—no reduction is possible.

There is no such thing as genetically constituted general intelligence since intelligence is a rational normative competence, the g factor is an epiphenomenal artifact of a historically contingent self-organizing cultural-cognitive ecology, and higher psychological functions originate as social relations mediated by cultural tools which only later appear individual. Hereditarianism tries to explain a normative status with causal mechanisms, a dynamic cultural artifact with a fixed trait, and an inter-mental function with intra-cranial genetics.

g is a third-person statistical construct. Intelligence, as a psychological trait, consists of first-person indexical cognitive acts. Category A – third-person, impersonal (g, PGS, allele frequencies, brain scans). Category B – first-person, subjective, experiential).

Genetic claims assert that differences in g (category A) are caused by differences in genes and that this then explains differences in intelligence (category B). For such claims to be valid, g (category A) must be identical to intelligence (category B). But g has no first-person phenomenology meaning no one experiences using g, while intelligence does. So g (category A) cannot be identical to intelligence (category B).

Thus, claiming genes cause differences in g which then explain group differences in intelligence commits a category error, since a statistical artifact is equated with a lived, psychological reality.

A natural-kind trait must be individuated independent of the measurement procedure. g is individuated only by the procedure (PC1 extracted from tests chosen for their intercorrelations). Therefore, g is not a natural-kind trait. Only natural kinds can plausibly be treated as biological traits. Thus, g is not a biological trait.

Combining this argument with the Kim-Kripke normativity argument shows that hereditarians don’t just reify a statistical abstraction, they try to reduce a normative category into a descriptive one.

Vygotsky’s social genesis of higher functions

Higher psychological functions originate as social relations mediated by cultural tools which only later appear individual. If hereditarianism is true, then higher psychological functions originate as intra-individual genetic endowments. A function cannot originate both as inter-mental social relations and as intra-individual genetic endowments. So hereditarianism is false.

Intelligence is not something a sole individual possesses—it is something a person achieves within a cultural-historical scaffold. Intelligence is not an individual possession that cab be ranked by genes, it is a first-person indexical act that is performed within, and made possible by, that social scaffold.

Ultimately, Vygotsky’s claim is ontological, not merely developmental. Higher mental functions are constituted by social interaction and cultural tools. Thus, their ontological origin cannot be genetic because the property isn’t intrinsic, it’s relational. No amount of intra-individual genetic variation can produce a relational property.

Possible counters

“We don’t need reduction, we only need prediction/causal inference. We’re only showing genes -> brains -> test scores.” If genes or polygenic scores causally explain the intentional-level fact that someone got question 27 right, there must be a strict law covering the relation. There is none. All they have is physical-physical causation—DNA -> neural firing -> finger movement. The normative fact that the movement was the correct one is never touched by any physical law.

“Intelligence is just “whatever enables success on complex cognitive tasks—we can functionalize it that way and avoid normativity.” This is the move that Heikenhimo (2008) takes out. Any causal-role description of “getting it right on complex tasks” obliterates the distinction between getting it right and merely producing behavior that happens to match. The normativity argument shows you can’t define “correct application” in purely causal terms without eliminativism or circularity.

“g is biologically real because it correlates with brain volume, reaction time, PGSs, etc.” Even if every physical variable perfectly correlated with getting every Raven item right, it still wouldn’t explain why one pattern is normatively correct and another isn’t. The normative status is anomalous and socially constituted. Correlation isn’t identity and identity is impossible.

“Heritability is just a population statistic.” Heritability presupposes that the trait is well-defined and additive in the relevant population. The Berka-Nash measurement objection shows that IQ (and any psychological trait) is not quantitatively-structured trait with a conjoint measurement structure. Without that, h2 is either undefined or meaningless.

Even then, the hereditarian can agree with the overall argument I’ve mounted here and say something like: “Psychometrics and behavioral genetics have replaced the folk notion of intelligence with a precise, operational successor concept: general cognitive ability as indexed by the first principle component of cognitive test variance. This successor concept is quantitative, additive, biologically real and has non-zero heritability. We aren’t measuring the irreducibly normative thing you’re talking about; we’re measuring something else that is useful and genetically influenced.” Unfortunately, this concept fails once you ask what justifies treating the first PC as a causal trait. As soon as you claim it causes anything at the intentional-level (higher g causes better reasoning, generic variance causes higher g which causes higher life success), they are back to needing psychophysical laws or a functional definition that bridges the normative gap. If they then retreat to pure physical prediction, they have then abandoned the claim that genes cause intelligence differences. Therefore, this concept is either covertly normative (and therefore irreducible), or purely descriptive/physical (therefore being irrelevant to intelligence.)

A successor concept can replace a folk concept if and only if it preserves the explanatorily relevant structure. But replacing “intelligence” with “PC1 of test performance” destroys the essential normative structure of the concept. Therefore, g cannot serve as a scientific successor to the concept of intelligence.

“We don’t need laws, identity, or functional definitions. Intelligence is a real pattern in the data. PGSs, brain volume, reaction time, educational attainment and job performance all compress onto a single and robust predictive dimension. That dimension is ontologically real in exactly the same way as temperature is real in statistical mechanics even before we had microphysical reduction. The heritability of the pattern is high. Therefore genes causally contribute to the pattern. g, the single latent variable, compresses performance across dozens of cognitive tests, predicts school grades, job performance, reaction time, brain size, PGSs with great accuracy. This compression is identical across countries, decades, and test batteries. So g is as real as temperature.” This “robust, predictive pattern” is real only as conformity to culturally dominant rule systems inside modern test-taking societies. The circularity of g still rears its head.

Conclusion

Hereditarianism rests on the unspoken assumption that general intelligence is a natural-kind, individual-level, biologically-caused property that can be lawfully tied to, or functionality defined in terms of, genes and brain states. Davidson shows there are no psychophysical laws; Kim-Kripke show even functional definitions are impossible; Kripke-Wittgenstein show that intelligence is irreducibly normative and holistic; Richardson/Vygotsky show that g is a cultural artifact and that higher mental faculties are born inter-mental;

Because IQ doesn’t measure any quantitatively-structured psychological trait (Berka-Nash), and no psychophysical laws exist (Davidson), the very notion of additive genetic variance contributing to variance in IQ is logically incoherent – h2 is therefore 0.

Hereditarianism requires general intelligence to be (1) a natural-kind trait located inside the skull (eg Jensen’s g), (2) quantitatively-structured so that genetic variance components are meaningful, (3) reducible—whether by strict laws or functional definition—to physical states that genes can modulate, and (4) the causal origin of correct rule-following on IQ tests. Every one of these requirements is logically impossible: no psychophysical laws exist (Davidson), no functional definitions of intentional states is possible (Heikenhimo), rule-following is irreducibly normative and socially constituted (Kripke-Wittgenstein), IQ lacks additive quantitative structure (Berka, Nash, Michell, Richardson) higher mental functions originate as social relations (Vygotsky).

Now I can say that: Intelligence is the dynamic capacity of individuals to engage effectively with their sociocultural environment, utilizing a diverse range of cognitive abilities (psychological tools), cultural tools, and social interactions, and realized through rule-governed pra gives that determine the correctness of reasoning, problem solving and concept application.

Differences in IQ, therefore, aren’t due to differences in genes/biology (no matter what the latest PGS/neuroimaging study tells you). They show an individual’s proximity to the culturally and socially defined practices on the test. So from a rule-following perspective, each test item has a normatively correct solution, determined by communal standards. So IQ scores show the extent to which someone has internalized the relevant, culturally-mediated rules, not a fixed, heritable mental trait.

So the object that hereditarians have been trying to measure and rank by race doesn’t and cannot exist. There is no remaining, respectable position for the hereditarian to turn to. They would all collapse into the same category error: trying to explain a normative, inter-mental historically contingent status with intra-cranial causation.

No future discovery—no better PGSs, no perfect brain scan, no new and improved test battery—can ever rescue the core hereditarian claim. Because the arguments here are conceptual. Hereditarianism is clearly a physicalist theory, but because physicalism cannot accommodate the normativity and rule following that constitute intelligence, the hereditarian position inherits physicalism’failure, making it untenable. Hereditarianism needs physicalism to be true. But since physicalism is false, so is hereditarianism.

(1) If hereditarianism is true then general intelligence must be a quantitatively-structured, individual-level, natural-kind trait that is either (a) linked by strict psychophysical laws or (b) functionally reducible to physical states genes can modulate.

(2) No such trait is possible since no psychophysical laws exist (Davidson), no functional reduction of intentional/normative states is possible (Kim-Kripke normativity argument), and rule-following correctness is irreducibly social and non-quantitative (Wittgenstein/Kripke, Berka, Nash, Michell, Richardson, Vygotsky).

(C) Therefore, hereditarianism is false.