Romain Brette
LightNews LightNews
romainbrette.bsky.social
Romain Brette
@romainbrette.bsky.social
520 followers 110 following 140 posts
Looking at protists with the eyes of a theoretical neuroscientist.
Looking at brains with the eyes of a protistologist.
(I also like axon initial segments)

Forthcoming book: The Brain, in Theory.

http://romainbrette.fr/
Posts Replies Media Videos
romainbrette.bsky.social
*par ex, le développement des caractères sexuels mâles du rat dépend du comportement dit de "léchage anogénital" de la mère. Du coup c'est "l'environnement" du fils, qui est en partie les gènes de la mère, donc en partie du fils, bref la distinction gènes/environnement n'a juste aucun sens.
July 23, 2025 at 6:59 AM
romainbrette.bsky.social
Here's another extract from my book, with a quote of Darwin:
The engineer starts from the desired functionality, then designs and tunes a structure that fulfills this functionality. In contrast, evolution is a blind historical process: a structure gets created by chance, and might then be conserved and possibly refined if it serves a useful functionality; or some existing structure might change and take on a new functionality. A central argument of Darwin in The Origin of Species (Darwin, 1859) is precisely that we find striking anatomical similarities across different species despite the diversity of lifestyles, which only makes sense if we assume a common ancestor: “What can be more curious than the hand of a man, formed for grasping, that of a mole for digging, the leg of the horse, the paddle of the porpoise, and the wing of the bat, should all be constructed on the same pattern, and should include the same bones, in the same relative positions?”.
May 28, 2025 at 7:36 AM
romainbrette.bsky.social
Against "reverse-engineering the brain":
(The Brain, in Theory, chapter 1)
Yet, if we were to explicitly ask a modern neuroscientist whether the brain is actually an engineered device, she would certainly strongly object. Brains are not the result of intelligent design. This is a religious view of life that has been discredited by Darwinism. Why then are we to “reverse-engineer” brains, if brains were not engineered in the first place? This terminology is typically excused by adding that brains are engineered by evolution, not by God. But Darwin’s insight is precisely that evolution is not a case of engineering. Engineering is the use of knowledge to solve technical problems. It presupposes an external mind that plans and assembles machines according to a preexisting goal. But evolution has no goals, plans or knowledge; in other words, it is not an engineer.
May 27, 2025 at 8:16 AM
romainbrette.bsky.social
Sure! Some context maybe (previous paragraph):
Remember that classical neurocomputational models of visual perception transform an input image into a mental symbol, such as a face label. This is because psychologically, we see a stable image and we are led to believe that there must be a stable physical entity in the brain that stands for that image, a “neural representation”, which is then further processed (Ramsey, 2007). But when we looked at the biology of seeing, we realized that there is nothing like an image in neurons of the retina: the visual field is dynamic and the activity of neurons is transient, made of electrical spikes. The stability of the visual world is a property of perceptual activity, not an intrinsic property of raw perceptual material.
May 22, 2025 at 4:13 PM
romainbrette.bsky.social
Samantha in the TV series Bewitched
May 22, 2025 at 3:12 PM
romainbrette.bsky.social
Why a mental state is not some kind of physical state.
(The Brain, in Theory, chapter 8)
Consider the following thought experiment inspired from the TV series Bewitched (Brette, 2019c). In that series, Samantha, the housewife, is actually a witch. She can twitch her nose and everyone freezes except her. Then she twitches her nose and everyone unfreezes, without noticing that anything happened. For them, time has effectively stopped. What happens during that time is not a continuous experience of the same mental state. Rather, there is no experience, no mental image at all, and it is as if time had jumped from one instant to another. This is because an experience is not a thing but an activity (Erdin and Bickhard, 2018; Zahnoun, 2020). Despite the common terminology, a “mental state” is nothing like a physical state.
May 22, 2025 at 3:12 PM
romainbrette.bsky.social
Information is the epistemic phlogiston of neuroscience.
(Chapter 6 of my book, "The Brain, in Theory")
Thus, in neuroscience, information is typically conceptualized as a sort of epistemic phlogiston: animals know because they have “information” in their brains, which enters through the eyes and moves up to higher cognitive centers. This is a fundamental misconception based on a reification of a cognitive process (Zahnoun, 2020). That is, what is actually a process (getting informed, understanding) is framed as the acquisition of some epistemic phlogiston that we call “information”.
May 16, 2025 at 6:47 AM
romainbrette.bsky.social
What is computation? ("The Brain, in Theory", chapter 4)
A computation is essentially an effective method in a formal domain (as opposed to a domain of interactions with the world), generally in the domain of numbers, which can be applied by a human with just pen and paper – we will see in chapter 8 the significance of this detail. Clearly, a recipe is not a computation. […] A computation operates on data (not pasta). It has an input and output, and it implements a particular mapping from input to output by a sequence of steps that belong to a finite set of elementary operations on symbols. The latter restriction is crucial for the notion of computation. Otherwise, the concept of computability would be totally meaningless. For example, Alan Turing demonstrated that there is no program that can take another program as input and calculate whether that program eventually stops: the halting problem is not computable. Obviously, this theorem would be false if we were allowed to choose any kind of instruction (choose as an instruction the mapping that assigns a stopping / non-stopping label to programs). Therefore, a process can only be meaningfully called “computation” when it is composed of a finite set of elementary instructions. Otherwise, the process is just a mapping between different kinds of data, not a computation.
May 15, 2025 at 9:14 AM
romainbrette.bsky.social
Shameless advertisement from my forthcoming book "The Brain, in Theory" :)
Chapter on "biological computers":
Thus, there are different computational concepts that need to be disentangled and discussed separately. First, a computer is a programmable machine (Brette, 2022a). As we have seen, living organisms are organized, but otherwise they are not machine-like. There is also no useful sense in which brains are programmable (4.1). But do brains compute? Here we need to be more explicit about what we mean by “compute” (Figure 4.1) (4.2). If we mean that some behavior can be described as programs in the broad sense of a sequence of predetermined actions, as in the “TV program”, then that is certainly often the case. If we mean that we can follow algorithms, in the broad sense of a sequence of fully specified elementary actions to solve a problem, then that is sometimes the case, as when we follow a cooking recipe. But if we mean computation in the sense of computer science, then we are referring to something much more particular, which operates in the formal domain with a finite set of rules. What we call “information processing”, the transformation of input data into output data, includes computation but also non-computational ways of solving data problems. When we examine the normal behavioral repertoire of organisms, we find that a large part is interactional and dynamic, including perceptual processes (4.3). Neither dynamics (4.4) nor interaction (4.5) can be meaningfully characterized as computation, unless by “computation” we mean “anything that solves a problem”. This undermines computational reductionism, the claim that all of cognition is ultimately decomposable into cognitive atoms, which are elementary computations. Nonetheless, an important part of behavior does not rely on interaction, and it is especially in those cases that neurons are said to “compute”. Unfortunately, we will see that the classical concept of “neural computation”, by which brain processes are said to implement a computational model, is flawed beyond repair (4.6).
May 14, 2025 at 6:19 AM
romainbrette.bsky.social
Why science is not reductionist, and never has been.
(my forthcoming book "The brain in theory", chapter 3)
Thus, in scientific practice, models are established from two types of constraints: the general laws of physics, the fundamental ones being microscopic, and constraints about the object of study, which operate at the scale of the object. Therefore, even when a macroscopic phenomenon is said to be “reduced” to fundamental laws, modeling practice is never entirely bottom-up: there are fundamental microscopic laws as well as macroscopic constraints. The very idea of reductionism is itself highly fallacious: of course, all physical phenomena cannot be reduced to the same few fundamental laws, since they are different phenomena. They can be reduced to the same general laws plus specific constraints. Often, these constraints are inferred from theories established at the scale of the object of interest.
April 18, 2025 at 7:57 AM
romainbrette.bsky.social
Celle-ci relève carrément de la psychiatrie. On parle bien de l'institution soupçonnée d'avoir bidonné les rapports d'évaluation pour fermer des formations de philo?
www.letudiant.fr/educpros/act...
April 7, 2025 at 3:19 PM
romainbrette.bsky.social
That is because Shannon information does not formalize the common sense of information as knowledge. It formalizes a communication situation, and the world does not communicate to the mind: the mind makes sense of the world. Building scientific laws is a better analogy.
(Excerpt from my book below)
March 20, 2025 at 10:41 AM
romainbrette.bsky.social
March 5, 2025 at 8:40 AM