Reconciling Psychoanalysis with Neuroscience

Far from the dogmatic standoffs and mutual rejections often seen between opposing camps of purists, there are real bridges and opportunities for fruitful cooperation between neuroscience and psychoanalysis.

A key condition for this dialogue is the redefinition of what should have always inspired both: the scientific method, understood both

• in its variants, as applied to natural and human sciences;
• and in its concern for demonstration and falsification, whether dealing with individual cases or general laws.

I. Dogmatic Confrontations

There has long been talk of a “violent debate” or even a “fratricidal battle” between those who see humans as mere biological machines and those who view them solely as beings of spirit and ideas.

On the neuroscience side, some reductionists cling to the idea of the “neural man”: that the architecture of the brain alone accounts for all psychic functioning. They claim, “The brain secretes thought as the liver secretes bile,” suggesting that chemical neurotransmission alone explains the entire spectrum of mental life.

This view echoes the fallacy of Pharaoh Psammetichus, who isolated infants from all human speech to discover the natural language of humanity, only to be met with incoherence. Likewise, rare cases of authentic “feral children” (such as Joseph Ursini) failed to speak, strongly suggesting that brain architecture alone cannot explain complex psychic functioning.

Philosophical materialists reject the existence of any immaterial principle. For them, the mind is merely the sum of physiological processes governed by physical laws.

This, however, is a fundamental error. Language, regardless of metaphysical interpretations, is not immaterial. It belongs to the physical world. Spoken language can be recorded, visualized, and acoustically analyzed, as I witnessed during my collaboration with phoneticians at the renowned Institut de Phonétique in Aix, now the Laboratoire Parole et Langage (LPL). Likewise, written language is visible, printable, and photographable. Admittedly, meaning itself cannot be embedded in the material trace, but the material trace remains real, especially when we observe its effects through brain imaging.

Eliminativism, on the other hand, holds that our everyday understanding of the mind is a radical error, and that neuroscience will eventually prove that mental states refer to nothing real. Some even argue that the concept of consciousness will be discarded entirely.

This question, especially as it relates to machine consciousness, is far from resolved.

Eliminativism has since been overtaken by computationalism, a theory that likens the mind to an information-processing system and compares thought to rule-based computation.

This model aligns more closely with reality. In artificial intelligence, the expert system illustrates this point. It comprises a “well-designed head” (the inference engine) and a “well-filled head” (a base of facts and rules), and although implemented through a machine, it exists meaningfully in the realm of logical programming languages (e.g., PROLOG).

A methodological detour can be illuminating here. I refer to five of the six sociological approaches identified by Jacques Herman in Les Langages de la Sociologie:

• the positivist paradigm,
• interpretive approaches,
• dialectical methods,
• functionalist frameworks,
• structuralist analysis.

Each of these perspectives deals with a different object (e.g., social fact, sociohistorical totality, human world, social system, cultural code) and reflects distinct cultural contexts, pilot sciences, and types of explanation. While Herman’s framework focuses on sociology, it can be extended to other human sciences and helps clarify how various research paradigms might be used to approach “Artificial Subjectivity.”

It’s worth remembering that Freud’s early psychoanalysis was firmly positivist. He employed biological metaphors and was inspired by physicist Ernst Mach’s energetic theories. Later developments, particularly Lacan’s “unconscious structured like a language,” distanced themselves from both positivism and interpretivism, engaging for a time with structuralism (guided by linguistics), before eventually taking other paths.

Let’s not forget, however, that reductionism exists within psychoanalysis too: some analysts cling to a disembodied psychic world, bordering on mysticism or pseudoscience.

II. Two Forms of Reductionism, Equally Futile

Should we turn instead to advocates of convergence between neuroscience and psychoanalysis? Many of these attempts are, in truth, pseudo-convergences.

• Take François Ansermet and Pierre Magistretti, who propose a reciprocal alliance based on neuroplasticity, where biology serves psychoanalysis and vice versa. Their goal is to “reintroduce the subject into biology.”
• Or the case of neuropsychoanalysis, which has been thoroughly critiqued by Laurent Vercueil.

III. Our Position

We propose that neuroscience and psychoanalysis explore distinct yet complementary domains through equally valid scientific methods:

• Neuroscience should remain focused on understanding brain function and on developing brain simulations (e.g., neural networks, hardware-based models).
• Psychoanalysis, for its part, should explore how language, external to the individual, shapes the psyche of the child and later the adult. This includes investigating Artificial Subjectivity as a complement to Artificial Intelligence: a way to simulate subjective processes through software.

This approach yields a spectrum of positions, including:

A. Revisiting Computationalism

This theory conceives of the mind as an information-processing system and compares thought to computation—more precisely, to the application of a set of rules. Computationalism does not claim that all thought can be reduced to such calculations, but rather that certain mental functions can be understood through this model. It represents a synthesis between intentional realism, which affirms the existence and causal efficacy of mental states (akin to interpretive approaches), and physicalism, which asserts that all existing entities are physical (a positivist stance).

Therefore, this theory is not inherently materialist: even if thought relies on a material substrate (the brain), it can be studied independently of that substrate. This stands in contrast to the kind of reductive materialism often found in neuroscience. After all, the same idea can be expressed through very different physical media—spoken aloud, written on paper, painted on a wall, or encoded in a computer.

In this sense, computationalism aligns with methodological behaviorism: unlike ontological behaviorism, it does not deny the existence of mental states.

As noted above, “The same idea can be expressed across different physical media (voice, paper, wall, computer, etc.).” It is precisely this premise that informs our work. Using an original method of discourse analysis (that is, a linguistic approach), called Analysis of Subjective Logics (ASL), we aim to describe—“on paper,” or more effectively through articles and simulation programs—the world of ideas that inhabits human thought. Not rational thought, but rather the realm of fantasy: seemingly irrational, yet governed by a logic we seek to reconstruct and test through simulation.

B. Vygotsky and the Social Genesis of Thought

Lev Vygotsky developed a theory of higher psychological functions using what he called the genetic method, which framed cognition as a “social history.” According to Leontiev’s theory of “decentering,” transmission is not only hereditary but also cultural. Intelligence, Vygotsky argued, develops through psychological tools found in the child’s environment, foremost among them, language.

Practical activity becomes progressively internalized as increasingly complex mental operations, thanks to words and the concepts they generate. The child’s so-called “egocentric” speech is in fact social in nature, eventually transforming into internal language in adulthood. Language is thus a necessary mediator in both the development and functioning of thought.

This approach parallels modern psychoanalysis, which also emphasizes language, temporality, and the role of others in shaping the self. Where we differ slightly is in emphasizing the decisive role of the family cell as the initial vector of cultural transmission, preceding broader societal influence.

What is often labeled “egocentric” language actually originates from the adult, who persuades the child that they have an ego. The transformation of external speech into inner language is something we will explore in detail in future articles. We fully support the idea that “language is a necessary mediator in the development and functioning of thought.” This is why, alongside psychoanalysis, we consider linguistics to be the second foundational pillar of Artificial Subjectivity.

C. Neuroscience supports this view

Neuroscience itself recognizes that higher brain functions require interaction with the world and with others. Neurons that are not activated early in life undergo attrition. Human development depends on engagement with language and culture.

We fully endorse this: human beings cannot develop without external interaction, especially through language and culture. A philosopher of science once said, “Half of man is language.” Without it, we remain biologically Cro-Magnon, endowed with advanced animal cognition, but not truly sapiens. We are little more than mammals, primates with potential, but no fully realized human subjectivity.

D. The Computer Analogy, Flawed but Enlightening

The mind is to the body what software is to hardware. A new computer is nearly empty; it can only perform tasks if it is loaded with diverse programs. Similarly, at birth, the body has minimal psychic functions. The complexity of the mind emerges from environmental input.

• A factory-fresh computer has only its circuitry. Identical machines will acquire different skills (word processing, drawing, music, etc.) based on the programs their users install.
• A newborn, free from hereditary or congenital issues, will develop different skills (language, abstract thinking, emotional regulation, personality structure, etc.) depending on the input, often unconscious, of their caretakers.

Maintaining or repairing the computer is the job of an engineer, not a programmer. Likewise, medical treatment of the body is distinct from understanding or treating the mind. That task falls to psychologists and psychoanalysts, who should not be conflated with medical doctors, except metaphorically.

Claiming that chemical mediators in the brain alone explain mental functioning is akin to saying that electricity explains software. Neural activity enables but does not constitute thought. EEG and brain imaging might show the effects of a mental program running, but the program itself comes from the outside and can be reshaped independently of its neural implementation.

There are, of course, limits to this analogy. As the saying goes, “Comparison is not proof.” Still, it helps delineate the scope of Artificial Subjectivity, which concerns only software, the verbal, subjective “program”, while biology focuses on hardware: the body and the brain.

IV. How to Collaborate: Dividing Complementary Tasks

There is a shared acknowledgment of determinism in both neuroscience and psychoanalysis. The latter postulates psychic determinism, reinforced by neuroscientific findings such as those from Benjamin Libet.

A. The Blind and the Lame (Inspired by Florian’s Fable)

• Modern science (in the Galilean tradition) seeks a logical-mathematical formulation of reality, grounded in empirical observation (Alexandre Koyré).
• Psychoanalytic discourse is a derivation of this modern scientific model. It emerged thanks to science but resists simplistic, unified explanations of human subjectivity. While the imaginary, the unconscious, and fantasy still pervade it, psychoanalysis is not a science, it is a de-imaginarizing discipline.
• Modern psychoanalysis has no fundamental criticism of the scientific method. It merely observes that science has had to turn its back on subjectivity to function, thus excluding it from its object of study by design.
• Science, in this sense, is “the blind man”: it advances by ignoring certain things, and it succeeds in doing so. Psychoanalysis, meanwhile, “sees” subjectivity but lacks methodological “legs” to walk on. Its practitioners often idolize founding figures and fail to renew its foundations. Today, psychoanalysis risks becoming “the lame man”, brilliant in insight, but stagnant in method.

Yet both science and psychoanalysis share important traits: neither is “whole,” neither offers ultimate meaning, both dissolve static notions of “being.” They oppose the imaginary. Sadly, they often behave like feuding siblings, when what is truly needed is negotiation and bridge-building.

(See my later article: “Psychoanalysis and Science: Reconciling the Estranged Sisters”)

We humbly propose a form of cooperation between “the blind man” and “the lame man.” This is the principle behind the bridges we call analysciences, such as the Analysis of Subjective Logics (ASL), which:

• Uses linguistic tools at the intersection of hard and human sciences;
• Applies these tools to psychoanalytic objects, such as fantasy;
• Extracts subjective logics, non-binary, non-dichotomous structures derived from the Greek logos, which shape both rational and irrational processes.

*** Irrational does not mean illogical. Far from it. ***

Science long ignored the unconscious. That is no longer the case, now that cognitive neuroscience has introduced the idea of the “cognitive unconscious.” However, this is not the same as the subjective unconscious described by psychoanalysis (see my lecture “Psychoanalysis and Propaganda”).

Some studies, such as those using subliminal perception to facilitate problem-solving, offer compelling evidence. Take, for example, Benjamin Libet’s 1983 experiment showing that brain activation precedes conscious decisions by several hundred milliseconds. More advanced versions from 2008 show preparatory brain activity 7 to 10 seconds before the subject makes a conscious choice.

In France, Lionel Naccache’s 2006 book Le Nouvel Inconscient (The New Unconscious) raises questions about the overlap between psychoanalytic and neurocognitive models. But his arguments are only partially convincing.

• He begins by honoring Freud but then claims that Freud’s unconscious merely misattributes functions that actually belong to consciousness.
• He denies repression, failing to consider that it may stem from externally introduced programs rather than internally generated cognitive circuits.
• Like the detectives in Poe’s The Purloined Letter, Naccache may simply be looking in the wrong place, his “four unconsciouses” fail to coincide with Freud’s.
• If Freud’s unconscious appears to follow conscious logic, it may be because the statements made consciously by caregivers are internalized and later influence the subject unconsciously.
• How does Naccache explain the return of forgotten memories under hypnosis or during analysis?
• How does he account for the immediate forgetting of dreams, often as elusive as post-hypnotic suggestions?

The subjective unconscious, deeply tied to language, operates on a different basis than the cognitive unconscious. We will begin, in a future article, to outline the similarities and differences between the two.

B. Criteria of Scientific Validity

1. The Scientific Method in Its Variants

It seems appropriate to reject two caricatured extremes:

• The imperialism of the hard sciences, which claim dominion over the human sciences through numerocentrism and naïve positivism;
• The vagueness, sometimes verging on autistic opacity, of those in the human sciences who reject any form of formalization.

Point 1: The Tyranny of Numbers

Statistics can be misleading—especially when applied to human language, which is not a one-to-one code. Two examples illustrate this problem: Egyptian hieroglyphs and the word “régime.”

Each hieroglyph can take on four distinct contextual values: figurative, symbolic (via metaphor, metonymy, etc.), phonetic, and determinative (a silent marker that semantically classifies the preceding signs when ambiguity arises).

Statistical analysis applied to ancient Egyptian texts, without distinguishing these contextual layers, fails to yield any meaningful insight. (For more on this, see my article “Translation and Interpretation.”)

The same applies to modern languages. Roughly 90% of vocabulary is polysemous—it only gains meaning in context. Take the phrase “Let’s change the regime!”—Is it referring to a political system? A dietary plan? A bunch of bananas? An engine’s operating mode? A grammatical construction? Counting the number of times “régime” appears without considering context renders the data virtually useless.

As Jean-Claude Milner rightly observed:
“We part ways with the widely held notion that science requires quantification. We argue instead: there is only science where mathematization is possible—and mathematization begins where literalization and blind functioning are present.”
(Milner, J.-C. (1989). Introduction à une science du langage. Seuil.)

Analysis of Subjective Logics (ASL) was developed with this principle in mind. It establishes a literal terminology (as found in phonology or Chomskyan syntax), which is by nature non-quantifiable. While it occasionally uses quantitative tools—such as counting semantic “atoms” and “molecules”—these are always interpreted case by case, never aggregated into statistical sums.

Point 2: Rethinking the Scientific “Fact”

In science, the notion of a “fact” deserves redefinition. Linguistics, for example, works with transcribed or recorded corpora—real, material datasets.

Linguistics may thus offer a shared foundation for bridging psychoanalysis and neurobiology. Psychoanalysts speak of the unconscious as structured like a language; neuroscientists cannot deny that language exists—and science itself operates through language.

Imagine a Huron encountering a running computer: they don’t need to know where the programs reside or how they operate in order to observe that they are functioning, use them, or question their logic. Likewise, linguistic analyses based on corpora work perfectly well—without needing to understand what happens inside the brain.

Logic-Based Analysis: The Gardin and Molino Model

Gardin and Molino proposed a model of logical analysis as rigorous as mathematical reasoning, involving two layers of validation:

• Internal validation of theoretical models and expert analyses;
• External validation through the creation of simulacra.

Analysis of Subjective Logics strives to implement this validation protocol in the field of human sciences:

• Internal validation occurs when we design programs that replicate ASL’s analysis on samples of discourse that reflect subjective rather than rational argumentation.
• External validation is achieved by creating simulacra of “subjective dialects”—virtually indistinguishable from those of native speakers—which elicit the same subjective reactions in interlocutors using other dialects.

Structuralism was abandoned prematurely. It should be rehabilitated—provided we strip away its fashionable, and often misleading, associations.

The structuralist approach bridges the divide between positivism (fact-seeking) and interpretive introspection. It posits a logic-based materiality in human discourse—whether from a social actor, speaker, or patient—independent of the truth-value of what is being said.

Jean-Claude Milner refers to this as extended Galileanism:

“In its own way, structuralist linguistics is a method of reducing sensory qualities. Natural languages only engage the sensory realm through phonetic form. And in that domain, the method yields clear results.

This represents a form of extended mathematization—rigorous, constrained, yet autonomous from conventional mathematics. In the 1950s, linguistics became as literal a science as algebra or logic, while remaining independent of both, and it achieved empirical success across all natural languages. It behaved like a true Galilean science. This extended Galileanism relied on a new kind of mathematics and extended it to novel objects. One of those objects was language—the very feature that separates humanity from the natural order.

Similarly, Levi-Straussian anthropology, using comparable methods on non-natural objects (such as kinship systems), produced exhaustive, accurate, and demonstrable descriptions of how they function. Lévi-Strauss found in linguistics a methodological and epistemological ally. On this foundation, linguistics and anthropology developed into a coherent movement of thought, whose methodological unity and epistemological value are beyond dispute.

That Lacan—whose relationship to Galileanism was foundational, and who approached his object more through culture than nature—should be counted among the structuralists is entirely logical.”

Milner’s eloquent defense of structuralism—as promising as it was, and abandoned too soon due to shifting intellectual fashions—speaks for itself. We fully endorse his position.

Extended Galileanism forms one of the pillars of Analysis of Subjective Logics and of our broader vision of Artificial Subjectivity. It is rooted in a form of “extended mathematization”—not quantitative, but literal.

2. From Particular Cases to General Laws

Exact sciences often accuse psychoanalysis of not being scientific, arguing that science must generalize. But inductive generalizations from statistics are not always useful, especially when language is involved.

Conversely, an exhaustive analysis of a single case, if communicable, can be just as valid as a collection of cases. The specificity of psychoanalytic casework has scientific merit in its own right.

3. Analysciences and the Analysis of Subjective Logics (ASL)

“Analyscience” is a term introduced by the creator of Analysis of Subjective Logics (ASL), Jean-Jacques Pinto, in 2008.

Still provisional, the concept refers to a hybrid discipline that bridges psychoanalysis and science. Its creation is grounded in the possibility of a dialogue—both epistemological and methodological—between modern science and psychoanalysis.

ASL could be considered a candidate for this label of analyscience. If we define it schematically as a micro-semantics of fantasy, fantasy itself meets three key criteria:

• It originates from prior clinical experience (analytic sessions);
• It has the beginnings of formalization—it can be linguistically defined;
• The fact that this concept subsumes a set of verbal occurrences can be demonstrated through ASL, whose material is visible and testable. Moreover, ASL’s analytic procedures are both manually reproducible and computationally simulable.

ASL thus enables us to analyze certain dogmatic postures mentioned earlier—particularly those underpinned by fantasies that can be modeled logically.

To conclude, we do not propose a confrontation between the hard sciences of nature and the so-called soft sciences of the human. Rather, we advocate a complementary association between the sciences of hardware and those of software, in order to better understand the dual poles of what defines the human interface—what we call the human condition:

• On one side: the brain as a biological machine (bio-computer)
• On the other: the verbal human software (verbware), subdivided into:

• cogware, which relates to cognitive identification
  • and subjware, which pertains to subjective identification.

This interfacing begins in childhood. It is the process of identification, with two facets:

• Cognitive identification (cogware),
• Subjective identification (subjware).

To study these processes through simulation, we can fabricate from scratch:

• Cogware modules, relevant to Artificial Intelligence, which simulate the results of cognitive identification—using expert systems, for instance, which differ from neural learning networks (which might be called interfacials!);
• And subjware modules, which inaugurate the field of Artificial Subjectivity by simulating the results of subjective identification.

Contrary to the binary opposition embraced by some positivists—and their paranormal-inclined opponents—the true structure is not twofold (rational vs. irrational) but threefold: rational, irrational, and logical. The logical (from the Greek logos) structures both the rational and the irrational, though each in its own way. And the logic of the irrational—that is the domain of psychoanalysis, when it chooses to be logical.

We extend an open invitation to all researchers animated by scientific curiosity to contribute to the development of these analysciences.

N.B.:

• Regarding our neologisms (bio-computer, verbware, cogware, and subjware), it should be noted that in our Artificial Subjectivity project, we have been developing subjware for several years using basic computational linguistics techniques. These methods are now gradually being supplemented and ultimately replaced by Generative Artificial Intelligence, which constitutes the third foundational pillar of our approach.
• The critique of the binary opposition between rational and irrational, and the introduction of a triadic model—rational / irrational / logical—were central themes of each of our three conferences held in 2023. The idea that “the logic of the irrational is psychoanalysis—when it chooses to be logical” is one of the conceptual cornerstones of Artificial Subjectivity.
• Finally, the reader will have understood that both analysciences and the Analysis of Subjective Logics are intrinsic and foundational to the Artificial Subjectivity project.