Plant Behaviour, Adaptive Plasticity and Evolution: Reassessing the Debate on Plant Intelligence

Ultimately, the study of plant memory helps to redefine the very meaning of adaptation. Adaptation no longer appears as a mere passive response to environmental stimuli, but as a dynamic process through which the organism continuously integrates perception, experience and the regulation of development. Plants thus emerge as highly complex biological systems, capable of using the past to modulate their relationship with the future. In this context, the concept of internal self-regulation takes on particular importance. Originally proposed as a biological theory by Maturana and Varela, the concept of autopoiesis has since been developed within the fields of philosophy of biology and systems theory. It provides a useful conceptual framework for interpreting plant self-regulation [42]. Plant behaviour cannot be interpreted merely as a series of episodic responses to environmental stimuli but must be understood as the result of continuous processes of dynamic regulation involving the entire organism. Every plant constantly maintains an extremely delicate internal balance, regulating growth, metabolism, resource allocation and stress response through sophisticated feedback systems. Light, water availability, temperature, nutrients and attacks by pathogens simultaneously influence the organism, which must integrate different and sometimes conflicting signals. Self-regulation represents the plant’s ability to coordinate this information whilst maintaining the functional stability of the biological system. In physiological terms, this process corresponds to a form of dynamic homeostasis, which is fundamental to the survival of plant organisms in variable environments [43]. Plant memory is not merely about the passive retention of traces of experience but constitutes an essential component of the processes of homeostasis and adaptation.

Previous experiences do alter the way in which the organism interprets subsequent stimuli, contributing to the regulation of future physiological responses. A significant example is stomatal regulation. Studies by Taiz and Zeiger show that the control of stomatal opening and closing is one of the main mechanisms through which plants regulate the balance between photosynthesis, transpiration and water conservation [7]. Stomata control gas exchange and water loss through transpiration. Under water stress conditions, plants can progressively modify stomatal sensitivity based on previous environmental experiences. This allows them to optimise the balance between carbon dioxide uptake and water conservation, thereby improving their ability to survive. Energy metabolism is also continuously self-regulated. During periods of nutrient or light scarcity, the plant redistributes its resources, slowing down certain growth processes and prioritising functions deemed essential. This capacity for internal reorganisation demonstrates that plant behaviour does not depend on isolated individual reactions, but on the coordinated interaction of multiple physiological systems. Self-regulation is particularly evident in the roots. They do not grow haphazardly in the soil, but constantly adjust their direction, growth rate and branching in response to available environmental information. Roots integrate chemical signals, changes in moisture levels, the presence of physical obstacles and competition with other organisms, dynamically regulating their own architecture. In this sense, the root system can be interpreted as a networked adaptive structure, capable of exploring the environment and continually modifying its strategies. Hormonal regulation also plays a central role in internal self-regulatory processes.

Plant hormonal networks represent extremely sophisticated integrative systems, capable of coordinating growth, development and responses to environmental stresses through feedback mechanisms [44]. Auxins, cytokinins, abscisic acid, gibberellins and ethylene form highly integrated signalling networks, through which the plant coordinates growth, defence and development. These networks operate through complex positive and negative feedback mechanisms. A local change can rapidly propagate throughout the entire organism, generating coordinated responses. This aspect once again highlights the distributed nature of plant organisation.

Unlike animals, plants do not possess a central organ responsible for controlling vital functions. Self-regulation arises instead from the continuous interaction between multiple components. From a theoretical point of view, this organisation is reminiscent of contemporary models of complex systems and autopoiesis. According to Capra and Luisi, living organisms must be interpreted as self-organising networks in which stability emerges from the continuous interaction between the components of the system [45]. The plant organism thus appears as a dynamic network capable of maintaining its own identity through continuous processes of self-regulation. Memory becomes an integral part of the organism’s ability to preserve internal stability whilst remaining open to environmental changes. Adaptation therefore does not consist of a simple mechanical reaction to external stimuli, but of a continuous negotiation between the environment, past experience and internal regulation. This perspective contributes to a profound redefinition of the very concept of plant behaviour. Plants no longer appear as rigidly programmed organisms, but as dynamic biological systems capable of integrating memory, perception and self-regulation into highly sophisticated adaptive strategies.

Within this framework, evolutionary theory is not an external addition, but the criterion that distinguishes the most convincing cases from the more speculative ones. If a behaviour is repeatable, mechanistically grounded, linked to differences in fitness and comparable across species or populations, then it can be treated as a trait subject to selection. When the terminology exceeds the empirical evidence—as in the attribution of consciousness or fully animal-like associative learning—the recent literature urges caution. This review does not argue for plant consciousness. Rather, it addresses whether certain plant behaviours satisfy the operational criteria for cognition, regardless of subjective experience. As persistence of a physiological modification is not sufficient to establish learning or cognitive memory, the above phenomena should be distinguished according to their empirical status and conceptual implications. Table 1 summarises this evidential hierarchy and indicates the appropriate terminology for each phenomenon. The table distinguishes the empirical status of selected plant phenomena from the terminology appropriate to their description and from the strength of the cognitive interpretation that they may support. Persistent physiological or developmental modification is not treated as sufficient evidence of cognition or learning. Stronger cognitive interpretations require additional evidence of context-sensitive integration, flexible response modulation, and the use of information acquired through previous interactions.

This hierarchy is important because persistence alone is not enough for cognition. A physiological state can retain the effects of previous environmental conditions without constituting learning. Therefore, vernalisation, dormancy, priming and epigenetic inheritance should not be indiscriminately grouped under cognitive memory. A stronger cognitive claim requires evidence that retained information modifies subsequent context-sensitive behaviour in a way that satisfies independently specified learning criteria.

Throughout this section, the terms “cognition” and “intelligence” are used in distinct senses. “Cognition” refers to the processes by which an organism integrates biologically relevant information and modulates its subsequent responses according to its state, previous interactions and environmental conditions. In contrast, intelligence is used more narrowly to denote flexible, adaptive problem solving when more than one possible response is available. According to this definition, intelligence may be considered a manifestation of cognition, but the two terms are not synonymous. Furthermore, neither cognition nor functional intelligence, as used here, implies phenomenal consciousness. Everything that has emerged and that I have discussed in the preceding paragraphs—namely, the fact that plants are undoubtedly highly plastic, sensitive and informationally integrated biological systems, and that they exhibit their own specific behaviour—has opened up the topic of plant intelligence. It must be made clear from the outset that the debate on plant intelligence does not centre solely on a biological question—whether plants learn, remember, make decisions or act—but on a question of the philosophy of biology: which concepts of cognition, mind, agency and behaviour can legitimately be extended beyond animals with a nervous system. In recent decades, the growing focus on plant behaviour has led many scholars to question whether organisms lacking a nervous system might exhibit forms of cognition, learning and decision-making [13,35,36]. The main difficulty stems from the fact that the concept of intelligence is historically and culturally associated with animals, and in particular with human beings.

The mechanisms involved in experience-dependent modifications of plant responses are heterogeneous and may include hormonal and calcium signalling, changes in gene expression, and transcriptional and metabolic regulation. In some cases, they may also include epigenetic modifications. Therefore, the conceptual question addressed here concerns not any single molecular mechanism, but rather whether the coordinated operation of such physiological and signalling processes can justify the use of cognitive terminology under specified conditions. The biological complexity of these processes is not enough on its own; the important issue is whether they support context-sensitive integration and the flexible modulation of subsequent responses based on information acquired through previous interactions with the environment. Traditionally, intelligence has been defined in relation to the presence of a brain, specialised neurons and centralised cognitive processes. Applying this category to plants implies a profound re-evaluation of traditional interpretative frameworks, given that for much of the history of modern science, the plant world has been described through a mechanistic paradigm. The historical lineage of those who advocate the idea that I can speak of plant intelligence at various levels can be traced back to Darwin’s statement, according to which the tip of a plant’s radicle acts like the brain in animals [5]. It must be said that this analogy does not, in itself, amount to modern evidence of plant cognition; nevertheless, it has shaped the imagination and vocabulary of contemporary debate. Proponents of a strong thesis interpret this Darwinian statement as proof that Darwin himself believed plants possessed plant intelligence.

Hence the idea that plants are intelligent in a literal sense because they integrate signals, display goal-directed plasticity, memory, learning and distributed forms of information without a centralised brain. Among the most influential figures in this field are Anthony Trewavas, František Baluška, Stefano Mancuso and, more radically, Monica Gagliano. They have contributed significantly to research on plant learning and consciousness.

In current terminology, plant intelligence and plant cognition are not synonymous. Intelligence is often used in a broad functional sense—the capacity to modify behaviour to improve survival and fitness under real-world conditions. Cognition is a more technical and theoretically charged term. For some, it requires representations or computation; for others, it can be understood in enactive, ecological or biogenic terms, or as a fine-grained concept comprising partial capacities. It is precisely this plurality of definitions that is one of the reasons why the debate persists [35]. The interpretation of these phenomena largely depends on the theoretical conception of cognition adopted. According to Anthony Trewavas, intelligence should be defined not on the basis of the presence of specific anatomical structures, but in functional terms [3,36]. An organism can be considered intelligent to the extent that it is capable of perceiving environmental information, integrating it, modifying its behaviour and solving adaptive problems. As I have seen in the previous paragraphs, plants appear to meet many of these criteria. They simultaneously perceive a wide range of stimuli, integrate information from different sources and modulate their own development in response to the ecological context. Root growth is a particularly significant example. Roots do not develop randomly in the soil, but explore the environment selectively, orienting themselves towards areas richer in water and nutrients and avoiding unfavourable conditions. Furthermore, Trewas’s thesis is explicitly anti-neurocentric. Intelligence would not, in principle, require neurons or a brain. The position taken by Mancuso and Baluska is stronger—and therefore also more controversial—as they use more explicit language.

Mancuso defended the programme of plant neurobiology as an integrated study of signalling, communication and plant behaviour, drawing on long-distance electrical signals, analogies with networks and even metaphors such as “plant synapses” or “root-brain” [10]. The heuristic power of this vocabulary has been considerable, although there is a danger of over-relying on analogies. Since plants do not possess a nervous centre, but the entire organism participates in information processing, according to Stefano Mancuso, such behaviour demonstrates the existence of forms of decentralised problem-solving [52]. This distributed organisation represents one of the most innovative elements of the contemporary debate. In higher animals, behaviour is generally coordinated by a central brain. Plants, on the other hand, function through networks in which every part of the organism contributes to overall regulation. This is what is meant by distributed intelligence. Cognition would not necessarily depend on the presence of neurons but could emerge from complex systems capable of integrating information and self-regulating. The theory of complex systems has played an important role in the development of these interpretations. According to Capra and Luisi, many biological systems exhibit emergent properties that cannot be explained solely through the analysis of their individual components [45]. Plants would therefore represent a form of cognitive organisation radically different from that of animals, but no less sophisticated for that.

The disputes seem to concern not so much the empirical evidence itself, but rather the interpretative and semantic framework through which I refer to that evidence. One example is the phenomenon of systemic signalling. In Arabidopsis thaliana, Toyota and colleagues have shown that glutamate acts as a wound signal and that GLR channels convert this signal into calcium waves that propagate to distant organs, inducing defensive responses. The recent literature confirms the importance of rapid electrical signals for the coordination of the entire plant. There is no substantial controversy regarding the phenomenon itself: the controversy rather concerns whether to describe it as information processing in a cognitive sense or as sophisticated but non-mental physiology [28]. There are many other important examples that support the idea that plants discriminate between complex environmental signals. Arabidopsis modifies its defences when exposed to vibrations produced by herbivore chewing, distinguishing them from other vibrations. Cuscuta pentagona uses volatile compounds to locate hosts and assess their attractiveness. Even in this case, however, the transition from selective sensitivity to cognition is not automatic. Much depends on how broad or restrictive the theory of cognition adopted is [53]. In this regard, an emblematic position is that of Monica Gagliano, with her studies on learning in Mimosa pudica, to which I have already referred in the context of the physiological memory of plants [54]. The researcher argued that repeated harmless falls induced habituation of the plant’s leaf-closing response, with a persistence that she interpreted as an elementary form of learning and memory. Thus, accepting this line of interpretation, the experiments on habituation would have shown that plants can permanently modify their behaviour on the basis of experience.

These results suggest that learning is not the exclusive preserve of nervous systems. Far more controversial is the case of associative learning in peas. In 2016, Gagliano and colleagues reported, in a Y-paradigm, that Pisum sativum could associate airflow with a future light position, directing its growth towards the neutral “cue” [50]. In 2020, Markel attempted a replication using larger, blinded samples and found no evidence of the effect. This was followed by comments and further replications, with disagreement over differences in protocol [51]. Philosophically, this episode shows just how much the field depends on methodological details, pre-registration, observer blindness and standards of replicability comparable to those in comparative psychology.

The debate on plant intelligence has taken on a strong philosophical and epistemological dimension. However, one of the main problems concerns the risk of anthropomorphism—that is, interpreting processes for which less mentalistic explanations exist as choices, intentions, altruism or motherhood. The recent critique of the personification of mycorrhizal networks and mother trees is instructive in this regard [55]. Numerous scholars argue that using terms such as intelligence, decision-making, or memory can be misleading, as these categories derive historically from human and animal experiences. For example, Alpi and other colleagues challenged the legitimacy of plant neurobiology, arguing that, in the absence of a brain and nerves, neurobiological terminology causes more confusion than theoretical progress. Plants do not possess neurons, synapses or brains. Consequently, attributing cognitive abilities to them risks confusing physiological phenomena with mental processes [56]. In turn, Taiz put forward an even more specific critique, arguing that the adaptive complexity of plants is not evidence of consciousness. Plant evolution has followed an alternative fitness pathway, based on genetically and epigenetically regulated plasticity rather than on subjective experience.

At the same time, I must also avoid falling into the opposite trap—that of reducing any observed flexibility to mere hard-wired routines, whilst ignoring the multi-scale integration of plant behaviour. The most promising positions today are those that do not seek to establish once and for all whether plants are cognitive but rather invite us to break the problem down into more circumscribed capacities: information integration, decision-making, memory, anticipation, selective sensitivity and sensorimotor coordination. This line of thought is represented above all by Paco Calvo, Miguel Segundo-Ortin, Jonny Lee, David Colaço and, in Italy, Margherita Bianchi [13,57]. These authors tend to view plant cognition as a borderline case useful for reformulating the concept of cognition in a biogenic, enactive, ecological or piecemeal sense. The debate thus functions as an exploratory conceptual hypothesis. The most appropriate course of action is always to operationalise before interpreting. If one is studying learning, one must specify what type of learning: habituation, sensitisation, classical conditioning, priming? If one is studying memory, one must establish the relevant temporal and mechanistic basis. If decision-making is invoked, one must define the alternatives, the available information, the metric of choice, and the control for simpler explanations. This is precisely the thrust of the recent methodological approach that links cognitive theory, the philosophy of science and comparative psychology in the study of plants. Furthermore, it is important to link the behavioural level to the mechanical level without reducing the former to the latter, by combining, for example, the phenomenology of behaviour, calcium imaging, electrophysiology, hormones, high-resolution phenotyping and modelling.

In the lexicon of the philosophy of biology, I need bridging explanations between emergent traits and underlying causal networks, not slogans or mere inventories of correlations.

Rather than asking whether plants are intelligent in an absolute sense, it is more productive to evaluate which concepts are justified by the available empirical evidence. While some plant phenomena are uncontroversially behavioural, others satisfy operational criteria for memory or adaptive information integration. Much stronger claims concerning cognition or intelligence, however, remain dependent upon the theoretical framework adopted. Therefore, the debate concerns not only biological evidence, but also the explanatory scope of the concepts used to interpret it. Therefore, the debate on plant intelligence should not be framed as a binary opposition between “plants are intelligent” and “plants are not intelligent”. Instead, it concerns the legitimacy of different conceptual vocabularies applied to increasingly sophisticated empirical evidence. Operational definitions, explicit theoretical commitments, and methodological transparency are therefore essential if future research is to progress beyond terminological disputes. In the present account, terms such as “root brain” and “plant synapse” have, at most, a heuristic value. While they may draw attention to functional similarities, such as distributed signal integration, long-distance coordination, and local information processing, they should not be interpreted as claims of anatomical or mechanistic homology with animal nervous systems. Furthermore, analogy alone has limited explanatory value. Where neuromolecular terminology does not yield experimentally distinguishable predictions beyond those generated by the established concepts of electrical signalling, hormonal regulation, and vascular communication, the latter terminology is scientifically preferable.

The argument developed in this review therefore favours a graded rather than a binary interpretation of plant capacities. Firstly, attributing behaviour to plants is well supported when behaviour is defined in terms of context-dependent modifications to growth, physiology, or interaction with the environment. Secondly, the term “intelligence” can be used in a limited functional sense to describe flexible, adaptive problem solving, but this usage should not be confused with psychological intelligence. Thirdly, cognition is a stronger and more controversial attribution. According to the non-representational account defended here, only cases involving integrated, context-sensitive, flexible regulation, where previous states or acquired information modify subsequent organism-level responses, are candidates for minimal cognition. Merely being adaptive does not satisfy these criteria for ordinary metabolism, homeostasis, developmental plasticity and physiological self-maintenance. Furthermore, none of the material reviewed here provides sufficient evidence to attribute phenomenal consciousness or subjective experience to plants. Therefore, the resulting position is neither that plants are intelligent in an animal-like sense, nor that all plant adaptation should be reclassified as cognition. A more defensible conclusion is that a subset of plant phenomena constitutes a legitimate test case for minimal, non-conscious, non-representational cognition. Whether individual phenomena cross that threshold must be established experimentally rather than inferred from biological complexity alone. Future progress will consequently require experiments designed not merely to demonstrate complex plant responses, but also to discriminate between competing explanations of these responses.

Priorities include:-Preregistered and adequately powered replications of controversial learning experiments;-Standardised experimental protocols;-Blinded behavioural assessment;-Explicit reporting of effect sizes and uncertainty;-Comparative studies across species and populations;-Experiments that directly contrast cognitive hypotheses with simpler physiological alternatives.

This approach would shift the focus of the debate from disputes over terminology to experimentally tractable questions concerning the organisation and evolution of adaptive control in living systems. The controversy surrounding plant intelligence ultimately does not arise from conflicting empirical observations, but from the different conceptual frameworks used to interpret them. The same experimental evidence may lead to different conclusions, depending on whether intelligence is defined as neural computation, adaptive problem solving, information integration or embodied cognition. Rather than asking whether plants are “really intelligent”, it is more productive to ask which definition of intelligence best captures the biological phenomena under investigation. From this perspective, the debate shifts from asking a simple yes-or-no question to conducting a conceptual analysis of the explanatory value of different theoretical frameworks. However, intelligence may still be used in a limited, functional sense to describe flexible, adaptive problem solving, provided this usage is clearly distinguished from stronger, more mentalistic conceptions involving conscious or representational capacities.