Towards the Idea of Molecular Brains

How can single cells without nervous systems perform complex behaviours such as habituation, associative learning and decision making, which are considered the hallmark of animals with a brain? Are there molecular systems that underlie cognitive properties equivalent to those of the brain? This review follows the development of the idea of molecular brains from Darwin’s “root brain hypothesis”, through bacterial chemotaxis, to the recent discovery of neuron-like r-protein networks in the ribosome. By combining a structural biology view with a Bayesian brain approach, this review explores the evolutionary labyrinth of information processing systems across scales. Ribosomal protein networks open a window into what were probably the earliest signalling systems to emerge before the radiation of the three kingdoms. While ribosomal networks are characterised by long-lasting interactions between their protein nodes, cell signalling networks are essentially based on transient interactions. As a corollary, while signals propagated in persistent networks may be ephemeral, networks whose interactions are transient constrain signals diffusing into the cytoplasm to be durable in time, such as post-translational modifications of proteins or second messenger synthesis. The duration and nature of the signals, in turn, implies different mechanisms for the integration of multiple signals and decision making. Evolution then reinvented networks with persistent interactions with the development of nervous systems in metazoans. Ribosomal protein networks and simple nervous systems display architectural and functional analogies whose comparison could suggest scale invariance in information processing. At the molecular level, the significant complexification of eukaryotic ribosomal protein networks is associated with a burst in the acquisition of new conserved aromatic amino acids.

Knowing that aromatic residues play a critical role in allosteric receptors and channels, this observation suggests a general role of π systems and their interactions with charged amino acids in multiple signal integration and information processing. We think that these findings may provide the molecular basis for designing future computers with organic processors.

At about the same time that Aristotle (384–322 BC) was elaborating perhaps the first organised reflections on “what is life?” in his work “Peri psyches—On the soul” [1], Zhuangzi (389–319 BC) was dreaming that he was a butterfly [2,3]. However, when he woke up, he wondered if it was not the butterfly that was dreaming of Zhuangzi? This dream was so lively that one still wonders if Zhuangzi really existed. However, the butterfly’s dream also transports us into a question that complements that of Aristotle: “what is reality, and how does human, butterflies and more generally, life sense it, build it and navigate between themselves, auto-stimulation and their perception of the universe. Intriguingly, these Western and Eastern thoughts also come together in the following strange convergence: in ancient Greek, psyches, “the soul”, also means “butterfly”.

While “this is a butterfly!” is the most common response on cards 1 and 5 of the Rorschach projective test, whose ink stains can reveal the deep structure of the human mind [4,5,6], monarch butterflies are also a gateway to quantum mechanics in biology. Thanks to their cryptochromes, which contain pairs of radicals, monarch butterflies perceive the Earth’s magnetic field and their wingbeats, can guide them over thousands of kilometres and bring us into the fascinating world of quantum biology [7,8,9]. The flapping of their wings that has also become a famous metaphor for chaos [10] could equally well illustrate the processes of cell signalling amplification, in which the weak stimulation of a receptor by a ligand or a photon, the flap of a butterfly’s wings in Brazil, can set off a cascade of molecular events that induce the overall change in the behaviour of an organism, a tornado in Texas. Thus, “butterfly effects” illustrate unexplored lands that touch both the essence of life and soul. First, they translate in a poetic way the evanescence of one of the fundamental properties of life, which still escapes the investigation of modern science: what animates it and does it have a will? This notion that Aristotle raised very early on by attributing a psychès, meaning a “soul” or “a form” not distinct from the body, to all living beings, be they plants or animals, is surprisingly topical in the light of a set of recent works on “consciousness” in single cells and organisms without a nervous system [11]. Second, the butterfly’s dream addresses the question of doubt and incertitude principles, where philosophy and quantum mechanic could meet in their fundamental question about our relationship with the reality and the limits of measurments to apprehend it.

Let us remember that much later in 1641, Descartes in his first Meditation will give us more or less the same dream as Zhuangzi and concluded: “All that I have received so far for the truest and most assured, I have learned from the senses, or through the senses: but I have sometimes felt that these senses were deceptive, and it is prudent never to rely entirely on those who have once deceived us” [12]. Finally, butterflies illustrate the volatile and moving world, woven of order and disorder where life likes to lodge itself for playing hide and seek with simplistic mechanistic principles.

Speaking about “brains”, even molecular ones, and “consciousness”, even cellular ones, are intended to examine what comes from oneself and what comes to us from the outside. Such precautions regarding reality are not only metaphysical and concern, as closely as possible, the nature of life and how its components perceive and respond to external or internal fluctuations, from the nano- to the macroscopic scales. In his seminal paper of 1995, Dennis Bray proposed that proteins can form circuits equivalent to the nervous system within cells [13]. However, he also suggested an equivalence between the “intrinsic activity” of the brains (as for example dreaming) and of the cells [14]: “both higher mammals and single cells (and by implication everything in between) generate movements in a proactive mode”. Supported by many experimental observations, this view considers that cells or organisms “continually rehearses possible future actions” that are “selected from an upwelling of spontaneous activity that serves to anticipate incoming stimuli” [14] (and references cited in). Can unicellular organisms, organelles and even macromolecular complexes such as ribosomes or stressosomes dream, be lured or confuse internal and external stimuli? According to a recent review by Mitchell and Lim, the answer would be yes: and cells would be prone to misperceptions, analogous to visual illusions, sometimes leading to incorrectly decoding input patterns of their environment [15]. Interestingly, these observations can also be extended to the molecular level. It is now admitted that, rather than being simple switches (ON/OFF), membrane receptors that pre-exist in a balance between active and inactive conformation can be activated spontaneously in the absence of a ligand [16,17].

More than a simple phenomenon of scale, this transposition could be the occasion to reconsider under another angle the biological macromolecules that “already” behave as complex systems [17,18,19,20,21,22]. Could the “emergence” phenomena from molecular and neural networks provide a kind of “scientific form” to the Aristotle psychès? Discussing “Molecular vitalism” 20 years ago [23] or “nano-intentionality” a little later [24] already incited to think differently about life and firstly questioned the pertinence of its hegemonic “machine metaphor”. In an odd paradox, Descartes, in spite of his awareness of the relativity of the interpretation of senses, is also at the origin of the “machine metaphor” in biology. Many years after his “meditations”, he proposed the analogy between human, animals or their organs with human-made machines that have imposed in a lasting way the mechanistic conceptualisation of life until today. However, this anthropomorphic conception of life starts to be seriously challenged, and it could be asked if the machine metaphor may hinder access to the essence and complexity of life.

In modern times of biology, the “machine metaphor” remains the pillar of the conception of life and its constituents. The ideas of “high performance” or “efficiency” of biological molecules optimised by selective forces are an integral part of the symbolic field of “a technological power”. In the collective imagination, machines are the fundamental actors of the industrial world, and they constitute the absolute references of reliability and efficiency of a productivist ideology. With “Pacific 231” and “Iron Foundry”, Arthur Honegger and Alexander Mosolov both composed symphonic pieces to the glory of machines, whether they were west or east of the Iron Curtain. As symbols of strength and power, for the societies in which science is produced (as by a machine), machines could have established their hegemony over the conceptualisation of life and its components. In his film “Modern Times”, Charlie Chaplin had grasped in all its depth the limits of this metaphor and the fate that the era would reserve for life and its “machine” metaphors. In a scene that is probably one of the most comic scenes in cinema, a machine is supposed to automatically feed a worker (The Tramp, C. Chaplin) so that he does not interrupt his work on the line. This infernal machine gets out of control and forcibly introduces all kinds of food into the mouth of the unfortunate “Tramp”. This tragi-comic scene shows the filmmaker’s ingenious intuition in the sciences of complexity: feeding without a feedback system is irremediably doomed to failure (Figure 1). Later, the film also illustrates in a masterly way how without the concertation of the mRNA and the three sites of the A, P and E tRNAs in the ribosome, the synthesis of proteins can very quickly become catastrophic (Figure 2).

Premonitory, would he anticipate the fateful fate that modern times have reserved for biodiversity? Feeding without feedback is indeed the antithesis of system biology, and such conceptualisation of life may severely hinder the understanding of the fundamental properties of its components. Thus following a strong selective pressure in the evolutionary history of ideas, the biological macromolecules are still today inexorably associated with molecular machines. We may wonder if it is not time to end this analogy, which dates back to the 17th century and which may bias or hinder our understanding of some of their properties that do not fit into this conceptual framework?

One sometimes finds in scientific literature: “Dancing in the cloud” [25], choreography [26,27] and symphonies [28] of life: rebel against anthropomorphic stereotypes, moving and escaping the dangers of reductionism, the metaphors coming from the artistic world seem much more relevant to conceptualise life and leave the way open for new ideas to emerge. For example, the choreographic metaphor seems to us to be much more appropriate than the machine metaphor for understanding how the concerted movements of molecules contribute to cellular life and its motility. However, although this metaphor has much richer heuristic virtues than machines for conceptualising life, it is little used compared to the machine (166 pubmed entries for “molecular choreography” versus 16,999 for “molecular machine”). Like cells, choreographic creation is based on the close relationship between the writing of a score (genotype), its interpretation (phenotype) and sometimes or even often, improvisation (epigenetics). In fact, if a choreography is generally written, the conventions of writing and transmission can be as numerous as the choreographers [29,30,31,32]. These different “codes” of expression correspond surprisingly to the great diversity of what is called cell signalling (see below), but they converge towards a single result: a global harmony of movements common to all cells. Although the gestures and movements of each dancer are precisely written and planned in time and space, they must constantly scrutinise each other discreetly to synchronise. The score alone is not enough to make a successful ballet. The interpretation, listening and constant adjustment of each dancer play a major role in the coherence, meaning and aesthetics of a choreography.

This notion is interesting when trying to understand the constant interactions between macromolecules that must constantly readjust in a fluctuating environment. It is fascinating, for example, to observe by molecular dynamics, the dance of two DNA double-helices closely intertwined, constantly looking at each other, attracting and repelling each other in turn in a kind of premise of love [33,34,35]: this scene is closer to a “pas de deux” than worm gear set. Yet, although machines and choreography are not very reconcilable, they are sometimes used to describe the same object. The ribosome, for example, may be an allosteric “molecular machine” [36] that can perform “the choreography of protein synthesis” [37]. Thus, see the molecules dance, contemplate the phenomena of emergence in the networks and their complex system behaviour’s [38] at all scales suggests that it may be time to escape from the machine metaphor in order to overcome the epistemological obstacles that may hinder the deep understanding of “what is life” [39].

Once freed from the conceptual stranglehold imposed by machine metaphors, today’s scientific context is conducive to re-examining both the “molecular vitalists” concepts and Denis Bray’s hypothesis [13]. On the one hand, many articles have recently highlighted the behavioural richness of microorganisms or plants lacking a nervous system [40] (see Section 2: “the psychic life of microorganisms”). They again raise the question of whether and how complex tasks can be performed without a nervous system. Cells contain a wide variety of receptors for virtually sensing all-physical and chemical stimuli (Section 3). These multiple signals are perceived, transmitted and integrated by the complex signalling networks to produce appropriate cellular “behavioural” responses. Cell signalling networks, which are now well characterised, are mostly based on the transient interactions of their components [38,41,42,43].

On the other hand, new types of molecular networks have been recently discovered in the ribosomes [44,45,46]. These “neuron-like” ribosomal protein (r-protein) networks display new properties compared to other “classical” biological networks [47]: they form mostly permanent connections through long disordered filaments and phylogenetically conserved tiny interfaces that have been compared to “molecular synapses”. These networks have evolved to optimise the interconnections between ribosome functional centres, thus presenting a “functional” analogy with simple sensorimotor networks [46] (Figure 2). They also probably use a new type of allosteric mechanism involving key aromatic amino acids. Conserved motifs formed by charged amino acids and π-systems are distributed along the whole networks and complexify during evolution. This suggested that charge transfer, propagation of electrostatic perturbation or even quantum phenomena may distribute signals throughout the network for synchronising ribosomal functional sites or even more complex tasks.

Because of their analogy with sensorimotor networks, r-protein networks initially suggested a metaphor that gradually evolved from “neuron-like circuits” at a molecular scale [44,45] to the one presented here: “the molecular brains”. However, the idea of molecular “proto-brains” has been previously proposed in the context of bacterial chemotaxis by Stock and his collaborators in the early 2000s [48,49,50]. These proto-brains, which consist of clusters of bacterial chemotactic receptors, have been shown to control bacterial motility in response to attractants or repellants. Already 20 before the proto-brains, the idea of “molecular brains” was germinating with the articles of Adler and Koshland proposing an analogy between chemotaxis and neurobiology [51,52]. More than a century ago, however, Charles Darwin and his son Francis already indulged in a similar analogy by extending the concept of the brain to plant roots: “It is hardly an exaggeration to say that the tip of the radicle thus endowed, and having the power of directing the movements of the adjoining parts, acts like the brain of one of the lower animals; the brain being seated within the anterior end of the body; receiving impressions from the sense- organs, and directing the several movements” [53,54].

However, the idea of “molecular brains” refers to the concept of the brain, and brains are still difficult to define… In a kind of “Unanswered Question” (Charles Ives, 1908), Vion-Dury and Mougin asked: “finally what is a brain?” in their paper “Neuroscience sans conscience n’est que ruine de l’âme” (a title that refers to François Rabelais’s novel “Pantagruel”). In their phenomenological approach, the authors conclude that modern neurosciences only give access to “fragments of experience, to blurred and perhaps false images of processes, to the distant shadows of the mind” [56]. Today, several ways of conceptualising the brain coexist more or less peacefully depending on whether one is in the camp of neurosciences [57,58], evolutionary biology [59,60], physiology, psychoanalysis [6], network science [61], physics [62,63], mathematics [64] or astrophysics [65]. “What is a brain?” would remain without a coherent answer for a long time to come, since the concept of the brain is so impalpable, and perhaps just as impalpable as the notion of the “psychès” and life.

When referring to neuronal networks and their connections, for example, synaptic, one usually uses the term connectome; this connectome, called anatomical connectome, is a structural description of the brain, and its connectivity properties can help exhibit central hubs of neurons for information integration, regions of highly interconnected neurons and important pathways of information [66]. Beyond anatomical connectome, networks can also be built from direct or indirect recording of neuronal activity by inferring statistical dependencies between the neurons (or collections of neurons); in this case, we refer to them as functional connectomes [61]. Several tools have been developed to study connectomes (centrality, rich club, small-worlds, graph similarities) [66,67]. We will, however, not detail them here. Behaviours can modify these structures of interactions, which lead to new ways to explore the brain-behaviour relation through structure-function relation.

The function of the brain is a global phenomenon that cannot be reduced to how its different areas function separately, and the relation structure/function would suffer from such reduction. In other words, wishing to find a relation between the function of an anatomical subpart of the brain and its local structure does not seem convincing, and one would rather like to understand how this subpart fits, with respect to its structure and function, in the whole brain. Until now, one would study the anatomical connectome as it is a global structure that one can have access to. However, the knowledge of the connectome is not sufficient for a global understanding of the function of the brain. For example, abnormalities of the brain connectome are known to be related to psychiatric disorders [68,69]; however, even though modifications of the connectome are observed, it would be difficult to interpret the disorder without an underlying model of how these dysfunctions appear [70,71]. The connectome is information that models have to take into account [72]; in other words, it is information that constrains the collection of relevant models so that they can, in turn, help understand the pathways for information integration.

Overall the connectome enables to exhibit interactions between relevant variables that come into play in the functioning of a “brain”; however, not all “brains” have a straightforward anatomical connectome [62]; for example, at the scale of the cell, there is no clear notion of an anatomical network of interactions between macromolecules and messengers that intervene in cell signalling although there is convincing evidence that this cell signalling is the physical support for information integration and decision making [73,74]. When looking for unified tools for describing these “brains”, one can turn to the more flexible notion of functional connectome can still be defined for these new “brains” as it relies on statistical inference; more generally, we propose, in the last section of this paper, the Bayesian brain as a unifying computational framework for what a “brain” is.

What is the nature of signals that flow through neural connectomes, and how do brains process this information? Two years after the storming of the Bastille, the publication in 1791 of “De viribus electricitatis in motu musculari commentarius” by Galvani was to bring about a lasting revolution in science, and in particular in neuroscience, by showing the role of electricity in nerve conduction and muscle contraction [75,76]. More than a century after the birth of electrophysiology, the development of increasingly sophisticated techniques has made it possible to analyse and understand the nature of nerve impulses in greater detail [77,78,79]. Today, all this knowledge forms the basis of computational or system neuroscience and current theories of neural coding. It is now well established that neurons communicate through a train of electric impulses (spikes) or action potentials [80]. The general class of models for describing how neurons generate spikes is called integrate-and-fire models, one of which is the Hodgkin-Huxley set of differential equations. Neurons can be classified into two types, inhibitory and excitatory. Their collective activity can be derived from the integrate-and-fire models of single neurons by weighting the connection of neurons with coefficients that depend on the strength of the inhibitory and excitatory effect of the synapses. This, in turn, could give some explanation for how decisions are made inside the brain (Chapter 16 in [79]). The activity of collections of interacting neurons is expected to represent information, in the sense of information theory [81], coming from the environment that would be relevant for decision making and functioning of the whole organism.

In their seminal paper, Adrian and Zotterman [82] demonstrated that during sensory perception, the intensity of a stimulus is “encoded” by a rate of nerve impulses over time and provided the first conceptual basis for neural coding and neural representations of the sensory worlds [83,84,85]. In addition, it is also thought that inter-spike intervals of individual neurons may also contain some information [86]. However, there still are debates on what is the relevant quantity extracted from the neural activity that is used for functioning purposes of the brain [78]: the “rate code” versus the temporal coding or “spike code” paradigms. For example, it has been shown that sensory information such as taste perceptions are encoded through temporal coding [87]. More recently, spike-timing code was shown to be also critical for motor coordination [88]. While the two views are not mutually exclusives, there are some cases where rate code is not enough for distinguishing external stimuli or where the presence or absence of one spike and its timing is sufficiently informative for inference on the source signal (Chapter 2 in [89]). This debate has been recently considered to be more of a methodological one, and the question of the relevance of the “neural code” metaphor was raised [90,91]. It has been recently found, however, that precise quantification of information available with the neural code could help rule out this debate [92] but it can be challenging experimentally [93], information is by nature dependent on the model of spike activity chosen (Chapter 3 [89]).

Of course, the “molecular brains” miss the extreme complexity of networks formed by neurons. However, studying simpler networks may have the advantage of allowing more direct access to the emergence phenomenon. The cellular stage of neural networks and the “black-box” of what occurs in the neurons considerably complicates the understanding of a central question in biology: how signals are propagated and integrated into a network, and how cognitive faculties such as decision making and learning can emerge from it? We infer here that if we look directly at the transfer and processing of information by connected molecules, it becomes theoretically possible to simplify the problem. This perhaps will help to extract what could represent universal principles of emergence in molecular or cellular brains, whatever their scales.

What are the differences and similarities between ribosomal protein networks and cellular signalling networks, and do they have functional and evolutionary relationships? And more generally, are the “informational” molecular networks (permanent or transient) related to nerve circuits? Is there an analogy or even homology between their “information processing” mechanisms at different scales? Or can we imagine that the nervous systems of metazoans and brains have borrowed from cellular networks the components and mechanisms at the origin of their evolution? Could it help to understand how information processing has evolved during major evolutionary transitions (i.e., from unicellular to multicellular life)? And finally, is there an evolutionary link between the mechanisms in molecular networks and in nervous systems? The nervous system of metazoans based on neural networks is well known to perform these sensorimotor functions. While the complexity of metazoan behaviour seems to be related to synaptic plasticity [94], where does the possibility of developing complex behaviour in molecular networks lie? How perception and motricity are integrated into beings without a nervous system remains a fundamental question that has not yet been completely resolved. These questions were asked more than a century ago by Alfred Binet in his book “the psychic life of microorganisms”.

Descartes, who introduced the “machine metaphor of life”, also contributed to the history of the nervous system. In his “Treatise on Man”, he proposed that “information” travels between the brain and the muscles via the nerves. Microscopy confirmed his first observations, and Leeuwenhoek (1695) showed that the nerves coming out of the brain were assembled in hollow fibrils [95]. However, it was only two centuries later that Cajal provided the basis for modern neuroscience, and Sherington would bring us into modern neurobiology, naming “synapse” what Cajal had previously identified as “nerve joint” [96].

Not only by his technical but also conceptual contributions, Santiago Ramon y Cajal was at the base of the neuron theory and modern neurosciences at the end of the 19th century [57,97,98]. By marking them with cell staining techniques developed by Golgi, he showed that all nervous systems are formed by networks of independent cells: the neurons. The meticulous observation of these networks allowed him to elaborate on his famous laws of optimisation of nervous systems: space, matter and time optimisation. In these networks, he also noted in which direction nerve impulses propagate. However, he has already perceived the complexity of the phenomena by noticing that the nervous system also influences the behaviour of peripheral receptors. Thus, he was able to develop an integrative approach that could be considered as the origin of systems biology, which built the foundations of neuroscience [99,100]. A key contribution to his understanding of the nervous system is to have integrated it into an evolutionary context [97,101]. This allowed understanding the general principles of their organisations, from the simplest to the most complex ones. Importantly, Cajal hypothesised the existence of a relationship between the architecture of neural networks and their functionalities, a working basis that will remain until today at the heart of the reflection in attempts to elucidate the functioning of nervous systems [96]. In addition, Cajal’s artistic sense, the quality of his representations and his wonder brought conceptual and technical rigour to science but also blurred the boundaries between science and the arts [102].

At the same time in France, Alfred Binet marvelled at another phenomenon that was in some ways close to that of Cajal but at a much smaller scale. “The psychic life of microorganisms” is the title of one of his intriguing works published in 1889 [103], where he describes the great behavioural richness of certain protists under the microscope. Binet, who was also at the origin of the famous IQ test (intelligence quotient) test, could not have foreseen that more than a century later, Michael Galperin would pass it to bacteria and classify them according to their behavioural types [104]. Binet defined two components in “the psychic life of microorganisms”: on the one hand, sensitivity, i.e., the action of the outside world felt by the organism, and on the other hand, motricity, the action of the microorganism on the outside world. Thus, a cell is “irritable because it has the property of responding with movements to the excitations it undergoes”: these are the bases of “cellular sensorimotricity”. He continues: if “psychic life is exercised by a nervous and by a muscular system in the great majority of pluricellular animals”, “it is not the same for microorganisms; most of them have neither a central nervous system nor sense organs; a few do not have locomotion organs”. In his work, he then lists all the organs of locomotion, sense organs that can be observed in protists and notes with relevance the analogies that they can represent with the corresponding organs in metazoans. He observes voluntary and involuntary movements, periodic movements and complex behaviours. Then he describes the complex and quite distinct behaviours that he has been able to observe in different species.

He also describes in passing the phototactic properties and sensitivity of bacteria to oxygen tension and about the microorganism nervous system he notes: “we have not found so far in a single proto-organism the slightest rudiment of a central nervous system”. Further on: “it has been said that if there is no anatomically differentiated nervous system in the lower organisms, it must be admitted that their cytoplasm contains a diffuse nervous system”. However, his work, which is of capital importance highlighting phenomena that are still largely unexplained, has long been forgotten. It is likely that the scientific context of the last century and probably “the machine metaphor of life” were not conducive to the birth and development of this theme, which shifts the question of what intelligence is at the molecular scale.

On the other side of the Atlantic, Jennings and Loeb followed in Binet’s footsteps by focusing on the “behaviour” of “lower organisms” [105,106]. Let us note in passing that this categorisation of life forms into “lower and higher organisms” was not conducive to the study of the behavioural complexity of these “lower life forms”. Just as it was unimaginable that continuity could be established between the animal and human world at the time of the publication of “The expression of the emotions in man and animal” [107], it was difficult to conceive at the time when scientists were still trying to demonstrate the superiority of the “white race” by cranial anthropometry [108] that a microbe could be intelligent. Nevertheless, Jennings, very early on, suggested an analogy between the microorganisms’ motility and the sensorimotricity of animals [105]. Half a century later, a decisive step was taken: Beatrice Gelber demonstrated that, like Pavlov’s dogs, paramecia could be conditioned and were therefore endowed with associative learning faculties [109]. Her work was fiercely contested and denigrated for decades, only to be rehabilitated recently in a fascinating review that reopens the question in all its mystery in the light of 21st-century biology [110]: how can beings without nervous systems perform complex behaviour? However, it was not until the 1980s that the behaviour of bacteria entered the disciplinary field of neurobiology [51,52].

However, the behaviour is as difficult to define as the brain and is considered as a kind of nebula [111,112,113] whose complexity often escapes the analytical methods of modern science. According to modern neuroscience, a behaviour is dictated by the central nervous system of animals and relies on the activity of neural networks. However, since Binet and Jennings, a growing body of research is converging on the idea that the ability to develop complex behaviours is not exclusive to beings with nervous systems, such as unicellular organisms or plants.

A fundamental and universal behaviour is, first of all, what we could call the “will to live”. Today, these “first wills” have spread, since the origins, to all scales of life and understanding “what bacteria want?” [114] would help to decipher the interplay between the “desire” to survive and the early stages of cell behaviour. One of the most basic forms of behaviour may be the distinct perception of the internal and external world, with the corollary ability to differentiate clearly between the self, other living beings and the physical world [115,116,117]. These faculties are shared at all scales of life: while the sense organs of metazoans are dedicated to the perception of stimuli of all kinds, unicellular beings have numerous sensors that perform an equivalent function but at the micro and nanoscopic scales. “Emotions” that depend on the perception of the world, its attraits or its dangers have probably been intimately interwoven with behaviours since the origin of life. For example, bacteria have soon developed large macromolecular assemblies called “stressosomes”, which are dedicated to reacting to any form of stress. Amusingly, a vast vocabulary from the emotional lexicon is also used to describe the behaviour of ribosomes. For example, one finds in the literature that they can be stressed, resting, taking a pose or even sometimes hibernating [118,119,120]. Moreover, “signalling” is also universal as the genetic code and has evolved from macromolecules to metazoans. Proteins communicate with each other through allosteric processes, using a complex “language” involving different types of mechanisms [121]. Within large macromolecular assemblies such as ribosomes, individual components constantly exchange signals with each other and ribosomes themselves can, by making specific contacts, exchange information [44].

Viruses can make decisions, such as choosing between lytic activity and lysogeny, by communicating through the emission of a chemical molecule, arbitrium [122,123,124,125,126,127]. In a similar way, quorum sensing allows bacteria and other microorganisms to inform each other about their density and trigger collective behaviour [128,129,130,131]. Very diverse modes of communication between cells have also been identified in unicellular beings and metazoans. The exchange of signals can be carried out by the emission of molecules that stimulate specific receptors in the host. However, inter or intracellular communication can also be established in a physical way by electrical, electromagnetic or acoustic waves or mechanical contacts [132,133,134,135,136,137,138,139,140,141,142,143].