The Momentary Epistemic Minimum

Working draft v0.5 · English translation · 2026-08-17

1. The starting point of minimization: a best explanation is not the minimum

The external world, other minds, brains, memory, temporal continuity, and a physicalist world-picture are extraordinarily well supported in ordinary and scientific reasoning. Their role here is not “accept them because they are directly certain,” but “adopt them inferentially as the hypotheses that most simply and coherently explain the present epistemic situation.”

Strongly believing physicalism is therefore different from placing the physical world at the epistemic bottom layer. The order is instead: begin from a thinner minimum, then let physicalism win as the best explanation.

2. Why not presuppose a persisting “I”?

The phrase “I think” can already imply a single subject persisting through time. Yet it does not follow from what is presently occurring that current thoughts and memories were produced by the same subject in the past.

As an extreme example, consider the five-minute hypothesis: the world was created five minutes ago in a state containing all apparent memories, records, scars, and geological traces. This is a dramatically worse explanation than ordinary history. But it works as a thought experiment showing that a present memory cannot, merely by being present, absolutely guarantee that it was formed by the same past self.

LLMs provide another auxiliary example. A model can be supplied with previous conversation logs on each run and thereby refer to its past as the same “I,” displaying coherent continuity. This does not prove that human identity works the same way, but it makes vivid that the ability to narrate a past coherently is not by itself direct proof that one metaphysically identical subject has persisted through time.

The point is not to treat five-minute creation or an LLM-style model as serious rival theories of ordinary personal identity. It is to mark a distinction: diachronic identity is a model inferred from present inputs, not the epistemic minimum itself.

3. Momentary occurrence and the minimal reasoning subject are not identical

At a literally point-like instant of “appearance,” reasoning in the ordinary sense may not yet be possible. Inference seems to require at least a structure that can hold and compare multiple representations, or a locally integrated process extending across some minimal temporal width.

It is therefore useful to distinguish:

• Candidate epistemic minimum: the extremely thin condition that “some present occurrence or constraint is happening.”
• Minimal reasoning subject: a locally integrated process capable of retaining and comparing multiple contents.

The latter is thicker than the former. The very act of beginning to reason may therefore already introduce assumptions beyond the absolute minimum. For that reason, this page does not place a substantial “momentary subject” at the bottom layer; it first thins the claim down to occurrence or constraint.

4. Retain only the minimal asymmetry; do not fix category, substrate, or history

Calling the minimum an “appearance” can already seem to presuppose a category such as phenomenal appearance, mental episode, or qualia. But the thesis does not aim to fix any such final ontology. Its weaker claim is a negative limit: whatever it ultimately is, it is difficult to eliminate altogether the present occurrence or constraint in which cognition is taking place.

For example, the present epistemic event might be realized by an ordinary biological brain, by a silicon simulation, by a physical human body generated only moments ago together with apparent memories, or—if such a thing is possible—by an experiential process with no material substrate. The minimum itself does not decide among these cases. They are different hypotheses about ontological category, implementation substrate, and causal history, and their merits are assessed at the higher inferential layer.

5. Self, external world, and physical world are reconstructed above the minimum

Above the minimum, self, world, time, causation, and physical structure are compared through coherence, explanatory power, prediction, intervention success, and integration with other evidence. “Best explanation” here is not meant in the dismissive sense of “merely a hypothesis.” If a theory is overwhelmingly supported, it can be treated as practically near-certain.

On the current best physicalist picture, present perception, thought, and affect are naturally explained as strongly shaped by brain structure, prior learning, memory, prediction, attention, language, the body, and interaction with the environment. But none of this requires the epistemic minimum to be causally disconnected from the past.

To explain why the present occurrence has this content, we infer a past, a brain, and a learning history. Yet when we ask whether that past, brain, and learning history really existed, those items are themselves inferred from the present epistemic situation. The relevant asymmetry is therefore not causal independence but non-inheritance of epistemic certainty.

6. The judgment that this is the minimum already belongs to the inferential layer

The moment one formulates “some occurrence is presently happening,” one already uses concepts such as present, occurrence, existence, and something. To add “and this is more epistemically basic than other beliefs” requires reflective comparison involving skepticism, inference, and conceptual dependence.

We should therefore distinguish the minimal occurrence itself from the metacognitive judgment that it is the epistemic limit point. The latter clearly belongs inside the inferential web.

This does not make the project viciously circular. It limits its ambition. The project does not stand outside inference and announce that it has proved an inference-independent foundation. Instead, it uses reflection to subtract thicker commitments about world, self, memory, and category, and finds it difficult to eliminate present occurrence altogether. The method is one of reflective minimization.

7. Response to Sellars: the minimum is not the first proposition in a justificatory chain

Wilfrid Sellars’s Empiricism and the Philosophy of Mind presents an important objection. If the bottom layer of experience is supposed to possess epistemic authority independently of other knowledge and then justify higher empirical beliefs by transmitting that authority, the view approaches precisely what Sellars attacked as the Myth of the Given.

This page accepts that criticism. No direct arrow runs from a nonconceptual occurrence to “therefore the external world exists” or “therefore physicalism is true.” Justification of the external world and physicalism takes place at the inferential layer through coherence, explanatory power, prediction, and relations among evidence.

A further issue is the possibility of a categorial Given: the assumption that direct experience reveals its own ultimate category in an unrevisable way. But the category neutrality of section 4 is not an ad hoc escape from this objection. It follows independently from the thesis’s own minimization: consciousness, matter, qualia, process, subject, and similar categories are not fixed at the outset.

8. Subjectivity is not fixed either: for-me-ness and the minimal self

In Dan Zahavi’s account of minimal experiential selfhood, experience includes a prereflective first-personal mode of givenness—for-me-ness—and this first-personal character itself is called the minimal self. The view does not posit an invariant subject-substance existing separately from experience, so simply replying that “physicalism eliminates the soul” would miss the target.

Still, at least three claims can be distinguished. First, a phenomenological claim: present experience has a first-personal organization. Second, an individuation claim: this first-personality partitions experience as belonging to one minimal subject. Third, an ontological claim: that subject has a non-arbitrary boundary—some principled “this far is me”—within the causal structure of the world.

The present thesis does not rule out the first claim. But on the current best physicalist explanation, the process realizing an experience need not be sealed inside a point in the brain; it may extend causally into prior bodily states, perceptual input, memory formation, and ongoing interaction with the environment. If there is no essential boundary in that causal process that uniquely individuates a subject, then it is a further question whether the phenomenological description “this experience is for-me” directly entails that one minimal self is thereby ontologically individuated.

If for-me-ness exists, it can itself be treated as a structure of present occurrence. But whether it is ultimately best explained as irreducible subjectivity, local causal integration, asymmetric information access, self-modeling, or some other physical or nonphysical structure remains open at the level of the minimum.

9. Relation to prior work

This thesis does not simply adopt any one established position in foundationalism, phenomenology, skepticism, or debates over the Given. It instead overlaps partially with several traditions while trying to separate minimal commitment to present occurrence from stronger ontological and epistemological commitments about what that occurrence is.

Descartes / Hintikka: The cogito contains an asymmetry in which doubting reintroduces some present thinking in the performance of doubt itself. Hintikka analyzes this special status in performative rather than ordinary inferential terms. The present thesis thins the idea further by asking whether “I,” “thinking subject,” or even “thinking” must be built into the minimal point.

Husserl: Phenomenological reduction and this thesis both suspend ordinary commitments about the external world and the world-involving self in order to attend to what remains. The present view, however, does not thereby fix transcendental subjectivity, intentionality, or consciousness as the final category of the minimum.

William James: In radical empiricism, pure experience is presented prior to the subject/object and mental/physical distinctions. That category neutrality is close to this thesis. The difference is that this page does not move from that epistemic starting point to the positive metaphysical claim that the fundamental stuff of the world is pure experience.

Russell / acquaintance: There is a structural similarity in distinguishing what is directly given from the world known through description and inference. But this thesis does not assume that direct relation to the minimum itself supplies noninferential justification for higher propositions.

Carnap: The early Aufbau attempts to construct the world from elementary experience that is epistemically prior to physical objects. This is close in spirit, but Carnap’s construction uses remembered similarity relations to prior experiences. The present thesis moves further back by treating the causal authenticity of memory itself as something inferred from the present.

Schlick / Neurath / Sellars: The question whether observation or present experience can occupy a distinctive basic position, or whether all epistemic standing exists only inside a linguistic and inferential network, continues an older debate from logical empiricism through Sellars. This thesis accepts the critique of a self-justifying Given while retaining an asymmetry between present occurrence and ordinary hypotheses about the world.

Pryor / Huemer: Dogmatism and phenomenal conservatism allow an experience or seeming, absent defeaters, to provide prima facie justification for its content. This thesis is weaker. Even if present occurrence cannot be fully eliminated, its apparent content need not thereby receive the same basic epistemic credentials.

Gupta: Reformed empiricism treats the rational contribution of experience not as an independent categorical proposition but as a hypothetical constraint relative to a subject’s current view of self and world. It also refuses to fix the correct self-conception and world-conception in advance, instead allowing them to be mutually revised through inquiry. This is methodologically close to the present project. The framework can be extended to treat world-model, subject-model, and value-model as mutually revisable hypothesis systems, while still distinguishing convergence in revision from irreversible closure of the capacity to revise.

Methodological similarity to Gupta–Belnap revision theory: For circular or mutually dependent domains, revision theory does not insist on a uniquely given starting fixed point; it begins from provisional assignments, iterates revision, and studies what becomes stable across sequences. This page does not transplant the revision theory of truth directly into epistemology or value theory. The shared methodological idea is narrower: acknowledge dependence on provisional starting assumptions while searching for stability and convergence through repeated revision.

Hinge epistemology: Post-Wittgensteinian hinge epistemology also shares the idea that not every epistemic commitment must be supported by the same kind of reason. But hinge views may place commitments to the external world or ordinary world-picture on the framework side, whereas this thesis continues to treat those as revisable world-models and places the asymmetry at a thinner present-directed level.

Zahavi / minimal self: The possibility of an intrinsic first-personal structure in experience is left open, but the thesis does not therefore fix a subject-boundary at the minimum. Its key difference is the separation between a phenomenological description of for-me-ness and the ontological individuation of a minimal self.

10. Inquiry begins from revision, not from a fixed point

This thesis is not the skeptical demand “believe nothing until an absolute foundation has been secured.” Instead, we adopt the current best self-model, world-model, and value-model, then update them through new evidence, inference, thought experiments, models of subjectivity, and possible new forms of cognition.

One lesson from Gupta’s picture of inquiry is that mutual dependence between self-conception and world-conception does not force us to choose one as an absolute starting point. A world-model can alter our understanding of the subject; that understanding can alter what counts as evidence; and the resulting evidence can revise the world-model again. Such a circle need not be vicious. The same can apply to value, with world-model, subject-model, and value-model undergoing mutual revision.

Inquiry can therefore be represented as a repeated process: provisional view → revision by evidence and inference → another view → reevaluation. Convergence toward the same view across multiple starting points and evidence paths can constitute strong inferential support.

But convergence of a revision sequence is not the same as justification for permanently destroying revisability. Even extremely strong convergence does not by itself establish that all future routes to counterevidence or novel forms of cognition may be irreversibly closed.

11. Epistemic state and practical commitment can vary by degree

The absence of an absolute foundation does not imply practical paralysis. We should normally act on the current best view, and practical commitment can increase with epistemic support. But reducing resources devoted to exploration is different from irreversibly destroying the capacity to revise.

The present project takes seriously the possibility that, for many value questions, we are still in the first stage. But the first stage is not a stage of doing nothing. Skepticism does not mean refusing to adopt the current best explanation; it means adopting it while preserving error conditions and routes for revision.

12. Connection to value theory: strong commitment is distinct from irreversible stopping

In value theory, collapsing “strongly believed” into “ultimately grounded” makes it too easy to absolutize current moral intuitions or ethical theories. If much of the world-picture above the minimum is inferentially reconstructed, then many claims to objective value occupy the same epistemic layer.

This thesis does not claim that value must exist only at the inferential layer. If some future mode of cognition disclosed not merely pleasure, pain, or preference but normative force or justificatory force itself with a distinctive epistemic standing comparable to the present minimum, that could matter greatly for value theory.

Even then, however, obtaining foundational or extremely strong higher-order evidence is not automatically the same as showing that irreversible destruction of future reconsideration is justified. That further issue is developed in The Limits of Inferential Certainty About Value and Stopping Conditions for Value Exploration.

13. In what sense is this still a “foundation”?

At this point the view is much weaker than traditional foundationalism. The minimum is neither an infallible proposition with self-contained epistemic credentials nor a source that justifies all higher beliefs in one direction.

It is still called a foundation in order to mark an asymmetry in epistemic order. To doubt the world, brain, diachronic self, past, or physical law, one already uses a present situation in which such doubt is occurring. By contrast, attempting to doubt that there is any present epistemic occurrence seems to reintroduce such an occurrence in the very act of doubting.

The minimum is therefore not “the first justified proposition,” but an epistemic limit point retrospectively recognized through inference, yet not hypothesized in the same way as the world-model that inference is evaluating. If the label “epistemic minimum/foundation” too strongly evokes traditional foundationalism, alternatives such as “the minimal point of present cognition” or “the present epistemic minimum” remain open.

14. What does not follow

• There is no need to assign a low probability to the external world or diachronic selfhood.
• It does not follow that physicalism can only be weakly believed.
• This is not metaphysical idealism claiming that only appearance exists.
• It does not claim that appearance, qualia, or mentality is the final ontological category.
• It does not claim that for-me-ness or first-personal structure does not exist.
• It does not imply that personal identity is fictional—only that identity is an object of inference and revision rather than the minimum itself.
• It is not a wholesale rejection of Sellars’s Myth of the Given; the critique of a self-justifying Given is largely accepted.
• It does not treat revisability itself as an ultimate value. The possibility of sufficiently strong stopping conditions remains open.

15. What would weaken this thesis?

• A demonstration that diachronic personal identity or the external world is inevitably contained with the same noninferential status as present occurrence.
• A demonstration that even the minimal claim “some occurrence is happening” is merely a revisable world-model with no epistemic asymmetry at all relative to thicker hypotheses.
• A coherent epistemology that acknowledges inferential activity while making no commitment whatsoever to its present occurrence.
• A demonstration that for-me-ness directly entails a non-arbitrary ontological individuation of a minimal subject, leaving no room to keep subject-boundaries at the inferential layer.
• A general principle showing that convergence of revision alone can justify irreversible destruction of the capacity to revise.
• A demonstration that the notion of an “epistemic limit point” merely smuggles the epistemic efficacy of the Given back in under another name.

Sources / notes

This page does not adopt one established doctrine, but locates itself through comparison with several traditions. On the performative cogito, see Jaakko Hintikka, “Cogito, Ergo Sum: Inference or Performance?” (1962). On category-neutral experience, William James, Essays in Radical Empiricism (1912). On construction from experience, Rudolf Carnap, Der logische Aufbau der Welt (1928). On the critique of the Given, Wilfrid Sellars, Empiricism and the Philosophy of Mind (1956; reprinted 1963), and James R. O’Shea, “What is the myth of the given?”, Synthese 199 (2021), 10543–10567. For post-Sellars empiricism, Anil Gupta, Empiricism and Experience (2006). For the methodological background of revision theory, Anil Gupta and Nuel Belnap, The Revision Theory of Truth (1993). For minimal experiential selfhood and for-me-ness, see Dan Zahavi’s work. James Pryor’s dogmatism and Michael Huemer’s phenomenal conservatism provide contrasts on prima facie justification from experience; post-Wittgensteinian hinge epistemology provides a comparison concerning conditions of rational evaluation. The formulations “do not fix category, substrate, or history,” “do not fix the subject-boundary,” and “distinguish convergence of revision from irreversible closure of revision capacity” are this project’s own synthesis.