The strongest philosophical position is not that the naturalistic fallacy has been disproved. It is that several quite different claims have been bundled under that label, and the strongest one survives while the broadest one does not.

The narrow principle remains formidable:

A purely descriptive fact, standing completely alone, cannot logically entail a categorical moral obligation.

But the much broader assertion sometimes attached to it,

“Facts about reality can never legitimately generate, modify, constrain, or help justify what we ought to do,”

is not an established philosophical truth. It is actively disputed, and quite a few serious philosophers have rejected versions of it. Contemporary moral naturalism is itself a major position in metaethics.

More importantly for our work, John Stewart's strongest argument does not actually require overthrowing Hume's is-ought principle. Stewart accepts its narrow version and works around it in a logically much safer way.

That may give us exactly the foundation we need for limited progressive evolution.

1. We first need to separate three ideas humans have inconveniently piled into one philosophical drawer

The term naturalistic fallacy is routinely used for things that are not identical.

Hume's is-ought problem concerns inference. On its standard interpretation, no normative conclusion follows deductively from premises containing nothing normative. If every premise says only what is, nothing in the premises supplies the logical content of ought. The Stanford Encyclopedia notes both the enormous influence of this interpretation and continuing controversy about exactly what Hume intended.

G. E. Moore's naturalistic fallacy is different. Moore argued that goodness cannot simply be defined as a natural property such as pleasure, evolutionary fitness, complexity, or survival. His famous open-question argument says that even after discovering that X has some natural property, it can remain a meaningful question whether X is good. Modern philosophical analysis points out that Moore's so-called fallacy is neither obviously a formal fallacy nor universally accepted as a mistake. Frankena and later philosophers argued that merely calling naturalistic definitions fallacious risks begging the very question being debated.

Then there is the much cruder appeal to nature:

Evolution produced X. Therefore X is good and humans should do X.

That inference is indefensible. Evolution produced cooperation, but it also produced predation, parasitism, deception, infanticide, lethal competition and assorted other charming accomplishments of life on Earth.

So these need to remain separate.

 

2. The strongest arguments supporting the is-ought restriction

After reviewing the arguments on both sides, I think the strongest case for maintaining a boundary looks like this:

This is why I would never put the following argument on the Universe Institute site:

Evolution tends under some conditions toward larger-scale cooperation; therefore humanity morally ought to pursue larger-scale cooperation.

Something important is missing.

The missing element is the normative bridge.

And this is where Stewart becomes much more interesting.

 

3. John Stewart's relevant work, and his strongest argument is better than “is becomes ought”

I searched Stewart's Evolutionary Manifesto site, its complete Manifesto, the publication collection associated with Human Superintelligence, his 2010 Foundations of Science paper The Meaning of Life in a Developing Universe, his 2024 book Human Superintelligence, and his 2026 BioSystems paper.

I did not find a separate major peer-reviewed Stewart paper devoted exclusively to the naturalistic fallacy. His most explicit treatment occurs in Part 4 of The Evolutionary Manifesto, is repeated and sharpened in his 2010 paper, and becomes particularly clear in Human Superintelligence. His current publications page links these works and his subsequent papers.

And Stewart begins by conceding the central point:

The fact that evolution favors something does not mean humans ought to favor it.

He then makes the crucial move:

His evolutionary oughts are derived from existing oughts or values plus relevant facts, rather than from facts alone.

That argument is explicit in The Evolutionary Manifesto.

In abstract form, Stewart's reasoning is:

Existing value or goal + relevant facts about reality → new practical ought consistent with the original value.

That is far stronger philosophically than claiming:

Fact → moral ought.

Stewart's survival example makes this unusually clear

His 2024 formulation is essentially:

If you want humanity to survive under specified circumstances, and reliable evidence establishes what humanity must do to survive those circumstances, then you ought to pursue those actions if you retain the original goal of human survival.

He expressly acknowledges that if a person has no preference whatsoever concerning humanity's survival, that particular ought does not bind them through this argument.

That is a hypothetical or conditional ought, not a categorical moral command.

And logically it is quite respectable.

Suppose:

V: We value the continued existence of a viable human civilization.

F1: Condition X will destroy or seriously undermine a viable civilization.

F2: Action Y is a necessary or highly effective way of preventing X.

Then:

O: Other things being equal, if we continue to value a viable civilization, we should undertake Y.

The scientific facts did not mysteriously manufacture the value.

But once the value exists, the facts can make previously optional behavior practically obligatory relative to that value.

That distinction is central.

 

4. Stewart has another good argument: there are no real “oughtless” humans

Stewart anticipates the objection that this merely pushes the problem backward forever:

Where did the first ought come from?

His response in Human Superintelligence is quite practical. Actual human beings already possess goals, preferences, commitments, needs and values. Ethical reasoning does not begin with a fictional organism containing no normative orientation whatsoever.

This does not logically prove that our existing values are morally correct.

But it defeats a weaker objection that conditional normative reasoning cannot get started until philosophy discovers an ultimate value axiom from nowhere.

Real deliberation starts with creatures who already care about things.

And evolution itself helps explain why organisms capable of action possess preferences, needs, goals, avoidance responses and goal-directed behavior. That biological explanation does not automatically justify every resulting preference, but it explains why practical reasoning never actually starts from a normative vacuum.

That point becomes important when we get to Bhaskar and Philippa Foot.

 

5. Where Stewart's case becomes weaker

I would not accept every step Stewart takes with equal confidence.

In the Manifesto, he proposes something approaching a universal fundamental value of favoring life over death and oblivion, and from there connects humanity's values to participation in future evolution.

That is considerably more vulnerable.

Someone can coherently value life while rejecting indefinite survival as the supreme value. People can sacrifice survival for justice, autonomy, dignity, protection of others or other commitments. Philippa Foot's work explicitly recognizes that human flourishing cannot simply be equated with survival and reproduction.

Likewise, Stewart's stronger scientific proposition that evolution has an overall trajectory toward increasing cooperative integration and evolvability remains a theoretical position, not an uncontested conclusion of evolutionary biology. His 2026 BioSystems paper explicitly advances that trajectory thesis.

For our limited progressive evolution framework, that is useful because it tells us where to draw the line:

Stewart's conditional fact + existing value reasoning is strong.

Stewart's empirical work on particular progressive evolutionary mechanisms deserves examination on its evidence.

His universe-scale trajectory and meaning claims should remain in our “contested interpretation/hypothesis” category rather than being treated as established science.

That separation actually strengthens Stewart's best argument.

 

6. Roy Bhaskar takes a more radical position

A small bibliographical correction first: the philosopher is Roy Bhaskar, and he was not principally a Cambridge philosopher. He studied at Balliol College, Oxford, later taught at Oxford and Edinburgh, and eventually held positions including World Scholar at the University of London Institute of Education. He was the founder of critical realism.

Bhaskar is highly relevant to what you are investigating because he went considerably further than Stewart.

In Scientific Realism and Human Emancipation, Bhaskar developed what he called explanatory critique, explicitly attempting to overcome rigid separations between fact and value, explanation and emancipation, and theory and practice.

The basic structure is roughly this.

Suppose social science discovers that people hold a false belief.

Then suppose research further discovers a social structure or causal mechanism that systematically generates or sustains that false belief.

Bhaskar argues that once we have reason to judge the belief false, we can, ceteris paribus, negatively evaluate the mechanism that systematically produces the falsity and gain a reason to transform it.

The Roy Bhaskar Centre's presentation stresses the crucial qualification: these normative implications are ceteris paribus, meaning “other relevant things being equal.” Open systems can contain countervailing conditions, competing needs or circumstances that prevent a general principle from determining what should be done in a particular case.

That qualification is extremely important for us.

Bhaskar is not offering:

Fact X → absolute universal ought Y under every possible condition.

He is much closer to:

Given a true causal explanation, relevant human interests or values, and absent overriding considerations, that explanation can give us a reason for action.

 

7. Bhaskar and Stewart overlap at an important point

Here is where I think Bhaskar can genuinely strengthen Stewart.

Stewart says:

Values + evolutionary facts can generate new conditional oughts.

Bhaskar says something closer to:

Facts and values are not always cleanly separable in actual social reality, particularly where facts concern human needs, false beliefs, causal constraints, agency and emancipation.

That is a deeper challenge to the rigid fact/value wall.

Critical realism treats reality as stratified and emergent. Social structures, persons, causal powers and human agency are real even though they cannot simply be reduced to elementary physics. Bhaskar also insists that the world is an open system, meaning causal mechanisms can be real without producing identical observable outcomes on every occasion.

That matters enormously to your question about scale.

We will return to it.

 

8. But Bhaskar has not “solved” the is-ought problem either

This is exactly where we need intellectual brakes rather than philosophical enthusiasm accelerating into a ditch.

Bhaskar's explanatory critique remains controversial even inside the critical-realist tradition.

A 2023 peer-reviewed analysis by William Hannegan examines Bhaskar's belief-based and need-based explanatory critiques and argues that both face a serious dilemma: either the normative conclusion does not validly follow, or a value has already been assumed in the premises. Hannegan concludes that Bhaskar's need-based version fails in its stronger form and that the belief-based version is insufficient by itself.

That objection actually pushes us back toward Stewart's safer formulation:

Don't pretend there is no normative premise. State it explicitly.

That is methodologically cleaner.

Instead of saying:

Fact → value

say:

explicit value + explanatory fact + causal relationship → conditional practical obligation.

Bhaskar still contributes greatly because he shows why the positivist picture of pristine value-free facts standing on one side of a philosophical wall and subjective values standing on the other is too simplistic.

But Stewart gives us the safer logical architecture.

 

9. Bhaskar is not alone. The absolute fact/value dichotomy has been attacked from several directions

This is important for your original hypothesis.

The proposition that facts and values must be absolutely isolated from one another is not an unquestioned consensus in contemporary philosophy.

Hilary Putnam explicitly attacked what he called the fact/value dichotomy, arguing that factual description and evaluation are often entangled rather than belonging to hermetically sealed conceptual universes. His argument makes particular use of “thick” concepts whose descriptive and evaluative dimensions are difficult to disentangle.

Andrew Sayer, working in a critical-realist tradition, similarly argues that some objective social descriptions require evaluative concepts and that framing all normativity as a simple “is versus ought” problem misrepresents actual social inquiry.

Svend Brinkmann has offered four arguments against an absolute naturalistic-fallacy thesis in psychology, including arguments from phenomenology, constitutive rules, human functions and descriptions whose factual and evaluative dimensions cannot easily be separated.

John Searle famously argued in 1964 that certain institutional facts can generate obligations. If someone validly makes a promise within a social practice whose constitutive rules create commitments, then facts concerning the promise can support a conclusion that the person ought to perform it. The success of Searle's deduction remains disputed, but it demonstrates how constitutive social rules complicate an absolute descriptive/normative separation.

Philippa Foot's mature ethical naturalism goes further into biology. Her concept of natural normativity argues that understanding organisms already involves concepts of function, need, flourishing, defect and successful operation. We can objectively say that a plant's diseased roots are defective relative to the life form without conducting a public referendum among the shrubbery. Foot then explores how this structure carries into human practical reasoning, while specifically refusing to reduce human good merely to survival or reproduction.

And directly within evolutionary theory, David Sloan Wilson, Eric Dietrich and Anne Clark argued in Biology & Philosophy that evolutionary psychologists sometimes invoke the naturalistic fallacy too broadly, using it to shut down legitimate investigation of ethical implications. Their point is not that whatever evolves is good. Rather, both ethical and unethical behavioral dispositions can have evolutionary explanations, and evolutionary findings can legitimately enter moral analysis.

Robert Richards has made an even stronger defense of evolutionary ethics, explicitly arguing that a revised evolutionary ethics can derive values from facts without committing what he considers a genuine fallacy. That remains more controversial than Stewart's conditional approach, but it proves that the philosophical question is very much alive rather than having been closed by Hume in 1739 and subsequently placed in a glass case.

 

10. Here are the conditions that qualify this issue

This is the most important result of this first research pass.

An evolutionary fact can legitimately contribute to a should or ought, without simply committing the traditional naturalistic fallacy, when the following conditions are satisfied:

1. A value, goal, need or normative commitment is explicitly present. We should not hide it. Examples might include human survival, minimizing unnecessary suffering, preserving agency, protecting future generations, sustaining ecological conditions necessary for civilization, or maintaining the capacity for continued learning and adaptation.
2. The evolutionary fact is sufficiently well established and narrowly specified. “Evolution favors cooperation” is hopelessly vague. “Under conditions A, B and C, mechanism M increases the stability of cooperation by reducing destructive within-group exploitation” is potentially useful.
3. There is a demonstrated causal or functional connection between the fact and the stated goal. Mere correlation, historical recurrence or evolutionary prevalence is insufficient.
4. The proposed action is actually relevant to achieving the goal. We need evidence that doing Y is necessary, strongly enabling or at least substantially more effective than plausible alternatives.
5. The level and system boundary are specified. What is beneficial for an individual may damage a group; what stabilizes a nation may harm humanity; what maximizes short-term human welfare may damage the biosphere. An “ought” cannot be transported casually between organizational levels.
6. Competing values and harms are considered. Increasing coordination, for example, could improve civilization's capacity to solve global problems while simultaneously threatening autonomy, diversity or freedom. A progressive capacity in one dimension can create regression in another.
7. The conclusion is defeasible and ceteris paribus. New facts, changed environments, unintended consequences or stronger conflicting obligations can alter the appropriate action. This is where Bhaskar's open-system reasoning is particularly useful.
8. We never substitute evolutionary success for moral goodness. Adaptive, successful, persistent, complex, cooperative, natural or evolutionarily favored are descriptive or functional properties. None is automatically synonymous with ethically good.

That gives us something far more useful than declaring either “Hume was right” or “Hume was wrong.”

 

11. There are actually four different kinds of “ought,” and this clears up much of the confusion

This distinction could become extremely useful in the Universe Institute section.

This distinction may be the conceptual key we have been missing.

When Stewart talks about what humanity “ought” to do to advance or survive future evolutionary conditions, his strongest arguments belong primarily in the second category.

They do not need to establish a categorical moral law written into the fabric of the universe.

That is an enormous simplification.

 

12. The subatomic-to-universe scale hypothesis needs one important modification

I do not think moving from subatomic scale to universal scale by itself changes Hume's logical problem.

An invalid inference does not become valid because the system becomes larger. Logic is annoyingly indifferent to grandeur.

But I think there is a much stronger version of your intuition.

As reality develops into increasingly organized levels, new kinds of entities and relationships emerge.

At the subatomic level, concepts such as health, function, suffering, obligation, agency, institution, truthfulness, justice and promise generally have no application.

At the biological level, however, we encounter organisms with functions, needs, regulatory systems, goals in the functional sense, viability conditions and failure conditions.

At the conscious-organism level we encounter preferences, suffering, anticipation and purposive action.

At the human social level we encounter reasons, commitments, norms, institutions, promises, laws, rights, reciprocal expectations and explicit values.

And at civilization scale we encounter systemic conditions necessary for maintaining civilization, ecological viability, knowledge systems, collective agency and intergenerational continuity.

That means the relationship between fact and normativity can become richer at higher levels not because Hume's logic expires at altitude, but because the ontology of the system contains new features.

Bhaskar's concept of emergence is especially relevant here. Critical realism explicitly rejects the assumption that everything meaningful about higher-level systems can be reduced to descriptions appropriate to lower-level physics.

This may be one of the deepest lines for our second-stage investigation.

 

13. Limited progressive evolution gives us a particularly good test case

Consider one of our limited progressive-evolution principles:

Under some conditions, the evolutionary transition to higher levels of cooperative organization requires mechanisms that control destructive internal competition while preserving enough diversity and autonomy for adaptation.

That is initially an empirical hypothesis.

It does not tell us that cooperation is morally good.

But now add:

We value maintaining a viable, adaptable, humane civilization capable of protecting individual dignity and surviving global challenges.

Then add an adequately demonstrated causal fact:

A civilization that cannot constrain particular forms of destructive internal competition cannot maintain the required large-scale cooperation.

We can now reach something resembling:

If we remain committed to a viable, humane and adaptable civilization, then, ceteris paribus, we should develop institutions that constrain those specific forms of destructive competition while preserving necessary autonomy, diversity and innovation.

That is an evolution-informed ought.

But the evolutionary fact did not magically become morality.

It specified what our existing values require under actual conditions.

I think this formulation is considerably stronger than simply saying an evolutionary “is” becomes an “ought.”

 

14. This also exposes an important distinction between “fact creates value” and “fact transforms what a value requires”

Suppose I value protecting someone's life.

Yesterday I believe a particular substance is harmless, so I have no reason to remove it.

Today reliable research tells me the substance will kill that person unless I remove it.

My value has not changed.

The facts changed what that existing value now requires me to do.

In ordinary language we might reasonably say:

“Because this substance is lethal, I should remove it.”

Philosophically, however, the complete argument is:

“Because I value protecting this person's life, and because this substance threatens that life, I should remove it.”

Humans omit the first premise because it is understood.

Stewart's argument is essentially that evolutionary knowledge can function in exactly this manner.

That is his strongest point.

 

15. This gives us a better formulation than “the naturalistic fallacy does not apply”

I would be very careful with that phrase on the Universe Institute website.

Instead, I propose a concept we can provisionally call:

Limited Evolutionary Normativity

A working definition:

Limited evolutionary normativity is the principle that reliable facts about particular evolutionary or developmental processes can legitimately help generate conditional, defeasible practical oughts when those facts are combined with explicit values, goals, needs or commitments and when a demonstrated causal relationship connects the relevant facts to achieving or protecting those values. Such oughts remain limited by context, system level, competing values, uncertainty, alternative means and changing conditions. Evolutionary facts alone do not establish categorical moral obligations or prove that whatever evolution produces is good.

That fits beautifully beside limited progressive evolution.

We would now have two mutually protecting concepts:

Limited Progressive Evolution prevents us from claiming that all evolution is universally progressive.

Limited Evolutionary Normativity prevents us from claiming that evolutionary facts automatically dictate morality.

And yet neither concept neuters evolutionary knowledge to the point where it becomes irrelevant to human decisions.

 

16. My present assessment of Stewart and Bhaskar

Stewart's most defensible insight is not that he has abolished the naturalistic fallacy. It is subtler:

Once humans possess values and purposes, scientific knowledge about evolutionary conditions can reveal consequences of those values that were previously invisible.

Bhaskar adds:

Facts, values, needs, functions, causal constraints and human agency are not always as cleanly separable in real social systems as positivist accounts suggest, and explanations of the mechanisms producing falsehood, suffering or blocked human capacities may legitimately have practical implications, subject to ceteris-paribus qualifications.

Putnam, Foot, Sayer, Searle, Brinkmann, Wilson and others show that Stewart and Bhaskar are hardly standing alone outside the philosophical city walls waving pitchforks at Hume. There is an extensive and serious literature challenging an absolute fact/value dichotomy.

But Hume's narrower challenge remains very powerful:

Where is your normative premise?

Our answer should not be to evade that question.

Our answer should be:

Here it is. We will state it explicitly.

That is, I think, the breakthrough for this project.

A provisional thesis for the Universe Institute

I would preserve this sentence for our next stage:

An evolutionary fact does not, by itself, create a categorical moral ought. But when people or societies hold an explicit value, goal, need or commitment, reliable evolutionary facts can reveal the conditions necessary to realize or protect that value. Under sufficiently established causal conditions, and subject to competing values, alternatives, uncertainty and ceteris-paribus qualifications, those facts can generate a rational, conditional and revisable practical ought.

And an even shorter version:

Facts alone do not tell us what to value. But facts can tell us what our values require.

That may be the strongest formulation I have found so far.

The next research level should now attack this thesis rather than merely decorate it. In particular, we should red-team Limited Evolutionary Normativity against the strongest objections from Humean philosophy, Moore, evolutionary debunking arguments, moral anti-realism, contemporary metaethics and critics of Bhaskar. Then we can determine exactly how much of this proposition survives before it becomes part of the Universe Institute's formal progressive-evolution framework.