The Characteristics of Animal Intelligence

Premise

Intelligence is difficult to define, as there are many functions in nature that appear to process information and aid in orientation. Likewise, the actions that follow—solving a problem, overcoming an obstacle, or moving closer to a goal—are equally diverse.

In the early 20th century, the French psychologist Alfred Binet (1857–1911) developed a test to assess intellectual abilities, focusing on faculties relevant to the school environment. In collaboration with the American psychologist Lewis Terman (1877–1956), the test was refined in 1916, leading to the concept of the intelligence quotient (IQ). Though later revised, this work laid the foundation for the Stanford-Binet Intelligence Scales, which hypothesised a single intelligence factor. Another significant measure, the Wechsler Adult Intelligence Scale (WAIS), was developed in 1939; it evaluates mnemonic and logical-mathematical skills, symbol-to-number associations, and tasks involving figure completion or reconstruction. However, both tests are influenced by an individual’s level of education. In response, efforts in the following decades sought to create “culture-free” tests, largely thanks to the work of the British psychologist Raymond Cattell (1905–1998).

From the 1960s onward, the cognitive revolution shifted the focus of intelligence research toward problem-solving and representational skills. Increasingly, intelligence came to be seen as the ability to process available information in pursuit of a goal. This perspective aligns with the definition proposed by  S. Legg and M. Hutter (2007, 12): “Intelligence measures an agent’s ability to achieve goals in a wide range of environments.” Intelligence, therefore, is understood as a systemic function that draws on various cognitive activities, including comprehension, associative reasoning, basic logical operations, planning, deduction, imagination, and creativity. Generally, a behavior is recognized as intelligent when it demonstrates certain key qualities, such as:

• • Context-adaptability – It is not merely repetitive but responds flexibly to the specific situation.
  • Goal-orientation – It has an intrinsic purpose, whether predefined (teleonomy) or emerging from subjective agency (teleology).
  • Result-based modulation – It incorporates feedback mechanisms to adjust actions based on their consequences.
  • Innovative capability – It goes beyond iterating pre-established patterns, always contributing something new.
  • Learning ability – It evolves over time by incorporating experience.
  • Problem-solving orientation – It aims to overcome obstacles, even by breaking existing schemes.

Intelligence is inherently elusive, both when it comes to defining its role in humans and even more so when comparing it across species. The challenge of providing a clear, official definition arises from its highly variable nature, as different cognitive functions take precedence in different contexts.

In 1904, the British psychologist Charles Spearman (1863–1945) attempted to define general intelligence through a g-factor. However, by the 1950s, Raymond Cattell introduced a distinction between fluid intelligence—the ability to solve novel problems—and crystallized intelligence—the ability to apply acquired knowledge and skills creatively. In 1965, the Czech psychologist Max Wertheimer (1880–1943), a key figure in Gestalt theory, further differentiated between logical intelligence, associated with analytical thinking, and creative intelligence, which fosters innovation and challenges conventional patterns—what is often called “thinking outside the box”. Later, Howard Gardner expanded the framework with his theory of multiple intelligences, while Daniel Goleman emphasized the role of emotional intelligence.

From this broader perspective, intelligence appears to be a common trait in the animal kingdom, manifesting through adaptive behaviors, flexibility in changing conditions, problem-solving abilities, understanding of complex situations, use of symbolic structures, memory retention, and the independent application of acquired knowledge and skills. Even among the simplest animals with a nervous system but no true brain, such as cnidarians (corals and jellyfish), we observe basic forms of intelligence. These creatures possess neural networks connected to sensory structures, enabling them to navigate their environment; they can orient themselves toward food sources, evade predators, seek out conspecifics to form shoals, and adjust their direction. We now know that they even exhibit some forms of learning, such as non-associative learning, which allows them to adjust their responses to stimuli through habituation or sensitization.

Even animals with a basic nerve center—essentially a simple brain—such as nematodes, demonstrate significant cognitive abilities. Caenorhabditis elegans, a widely studied model organism, exhibits advanced information-processing skills, including associative learning, exploratory behavior, and decision-making. It can distinguish between reward and punishment signals presented simultaneously, integrate information, and retain it over extended periods. Among flatworms, Dugesia japonica displays an extraordinary ability: when cut in half, it not only regenerates its nerve center but also retains memories from the original brain. This suggests that memory in these organisms is distributed throughout the body (S. Ginsburg & E. Jablonka, 2023).

Among vertebrates, fish also exhibit remarkable cognitive abilities. They can navigate mazes, use tools, and distinguish between subtle visual and auditory patterns. The white-spotted pufferfish (Arothron hispidus), for example, constructs intricate, radially symmetrical nests decorated with shells to attract females, who lay their eggs at the center. Fish have also demonstrated parental care, various forms of memory, social behaviors involving cooperation and communication, the ability to make choices between different options, and the strategic use of available resources. Studies on the anglerfish (Lophius piscatorius) have demonstrated its ability for associative learning and emotional anticipation in response to aversive stimuli.

Amphibians can be seen as a transitional group, exhibiting cognitive traits that become more pronounced in reptiles, mammals, and birds. Some urodeles, such as the red-backed salamander (Plethodon cinereus), have demonstrated decision-making abilities—when presented with two tubes, one containing two prey and the other three, they consistently choose the latter. Similarly, the emerald toad (Bufotes viridis) exhibits remarkable memory, particularly in orienting toward natal ponds during the breeding season, recognizing targets, and marking negative experiences.

Invertebrates also exhibit impressive cognitive abilities, as demonstrated by hymenopteran insects. Social learning, a key indicator of intelligence, has been observed in bumblebees (Bombus terrestris), which can learn tasks by watching trained individuals solve the same problem. Research at Queen Mary University of London has further shown that bumblebees engage in play and can form mental images based on spatial references. Additionally, early studies on cognitive maps were successfully applied to bees and other social hymenoptera, revealing their ability to navigate, recognize shapes and colors, memorize faces, and engage in complex symbolic communication.

It must be noted that these discoveries have gained traction in recent decades, as scientific perspectives have progressively moved beyond anthropocentric biases that once considered intellectual capacity exclusive to humans. This perspective gradually gave way to a more flexible approach, initially recognizing intellectual capacities in animals most similar to us, such as great apes. Over time, this recognition expanded to include mammals and birds, broadening the concept of cognition while still keeping our own species at the center. Today, however, the notion of intelligence is increasingly seen as plural. Studies on cephalopods, for example, reveal highly complex cognitive abilities in creatures far removed from the evolutionary lineage that led from reptiles to mammals and birds. These animals demonstrate problem-solving skills, the ability to learn new strategies, tool use, and even signs of dreaming.

Shedding our anthropocentric biases, we are left to ponder the question posed by ethologist Frans de Waal: Are We Smart Enough to Know How Smart Animals Are? (2016).

The Dichotomies of Intelligence

Examining how intelligence has historically been understood reveals a recurring dichotomy. On the one hand, the extensive conception of intelligence—a view that attributes intelligence to any entity capable of processing information appropriately, displaying goal-directed behavior, and adapting through self-regulation in response to circumstances. On the other, the intensive conception. This perspective reserves intellectual capacities for select entities, primarily humans, but also mammals and birds. It emphasizes not only the criteria mentioned above but also and especially the presence of imagination, creativity, problem-solving, and inferential reasoning. These two approaches are fundamentally opposed. The extensive view recognizes intelligence in any instance where an entity exhibits adaptive, flexible, goal-oriented, and self-regulated behavior. In contrast, the intensive view defines intelligence not merely as problem-solving but as intus legere—the ability to see beyond appearances, imagine what does not yet exist, create novelty, and, crucially, formulate problems rather than just solve them. Under this second perspective, even so-called artificial intelligence would not qualify as true intelligence, as it operates through computational and configurational processes rather than engaging with problems in an autonomous, creative way.

Another key distinction in understanding intelligence lies in the dualism between a quantitative and a qualitative view of it. The quantitative conception sees intelligence as a matter of degree and potential, akin to IQ measurements. This approach ranks intelligence on a gradient, often defining the most intelligent animals as those that most closely resemble humans. The qualitative view of intelligence, instead, rejects the idea of measuring it in absolute terms. Instead, it views intelligence as an emergent function that manifests differently depending on various cognitive faculties. Under this perspective, one individual may excel in thinking outside the box, while another demonstrates exceptional creativity—aligning with Cattell’s distinction between fluid and crystallized intelligence. The qualitative view reaches its highest expression in humans according to Gardner’s theory of multiple intelligences.

Regarding the two dichotomies discussed, I want to clarify my position from the outset. I favor an intensive conception of intelligence—albeit in a moderate form—that emphasizes the problem-solving and creative dimensions of intellectual activity. And when considering different species, I believe intelligence must be understood qualitatively, as it is one of many functions of an organism, inherently shaped by specialization.

Let’s begin with the extensive conception of intelligence, which considers all entities and performances that exhibit goal-directed behavior, context sensitivity, and self-regulation. The simplest example is negative feedback, a mechanism that allows a system to maintain a certain range—like a thermostat that regulates temperature based on its effects. Though it may seem as if the thermostat “knows” when to produce or stop heat, it is merely executing a self-regulating feedback loop. Similar homeodynamic mechanisms govern many internal regulatory functions in living organisms. These include temperature control, solute concentration balance, pH regulation, and endocrine processes—such as hypothalamic releasing factors, pituitary tropins, and hormone release—ensuring the body’s stability through negative feedback. We also observe positive feedback mechanisms, where a product stimulates its own increased production. These autocatalytic processes drive growth, organ reinforcement, and even degenerative phenomena. A striking example is the construction of a termite mound, which relies on positive feedback: each time a termite deposits a pellet of mud, it releases an engagement pheromone. The more pellets accumulated, the stronger the pheromone gradient, reinforcing the building process.

Another intelligent mechanism—one that generates increasing complexity—is the threshold system, which integrates both negative and positive feedback. In these systems, negative feedback operates below a certain threshold, maintaining equilibrium, while positive feedback takes over above the threshold, triggering transformation. A classic example is found in ecological climax systems: under normal conditions, CO₂ emitted by animals is absorbed by plants, keeping the system in dynamic equilibrium through negative feedback. However, if CO₂ levels rise beyond a critical threshold, positive feedback mechanisms accelerate climate change. The greenhouse effect increases temperatures, reducing the ability of plants and photosynthetic microorganisms to absorb CO₂; higher temperatures then also trigger wildfires, releasing even more CO₂. This shift from stabilizing negative feedback to runaway positive feedback leads to catastrophic system transformation.

The use of reference gradients to regulate communication and structural organization was first described by the French biologist Pierre-Paul Grassé (1959) as stigmergy. In termites, for instance, the size of the queen’s chamber is determined by the pheromone gradient emitted by her body. As the queen grows, she expands this gradient outward, signaling workers to enlarge the chamber accordingly. These examples illustrate the many modes of self-regulation in complex systems.

Indeed, a structure can be directed toward a particular outcome without having an inherent purpose, much like a river flowing toward its mouth or a chain of dominoes creating intricate patterns. In this sense, another form of intelligence is found in teleonomic or functional cascade mechanisms—processes structured so that their outcomes are already embedded in the system’s initial organization. Traditional algorithms operate in this way, consisting of a set of detailed instructions that, starting from a given point, lead to a predetermined result. However, an algorithm clearly does not achieve this because the system has an intrinsic purpose but rather because: i) it follows a defined sequence of instructions, where each step must be completed before proceeding to the next; and ii) it is deterministic and unambiguous, meaning that identical input data will always yield the same result. The algorithm is a fundamental concept in computer science, but similar performative intelligence can be observed in nature, such as in embryogenesis or certain choreographed movements.

The term teleonomy was popularized by the biologist Jacques Monod in 1970 to describe processes directed toward a final outcome, with their predisposition already shaped by natural selection. It is a closed mechanism, which differs from true goal-directed action, and closely resembles the domino effect of Washburn’s chain (S-R/S-R/S), where each response acts as a stimulus for the next—an explanation proposed by behaviorists for complex behavior. Essentially, this mechanism mirrors that of an algorithm: a predetermined sequence of operations leading to a specific result. For a long time, innate behavior was understood through this teleonomic model, where animals were thought to follow rigid, preordained patterns.

That, however, was a wrong assumption. We now know that even bacteria can monitor their environment, adapt their behavior, and influence their surroundings. They exchange genetic material and, through quorum sensing (QS), communicate and regulate gene transcription. This mechanism involves the release of specific molecules that accumulate in proportion to bacterial concentration. Once a threshold level is reached, it triggers a coordinated behavioral shift. For example, Vibrio fischeri emits light only when its population surpasses a certain density. Bacteria use QS to facilitate colonization, cooperate, and develop new states and functions.

Plants also exhibit a form of intelligence, demonstrated by their ability to learn, remember, and adjust responses based on past experiences. Many behaviors typically associated with animals—such as resource competition, adaptation to environmental changes, cost-benefit analysis, mutual aid, and active environmental modification—are already present in the plant world. Plants communicate, particularly through mycorrhizal networks (S. Simard, 2018), forming an underground system often referred to as the Wood Wide Web. This network not only facilitates information exchange but also enables complex, context-dependent coordination of growth and development, including highly sophisticated forms of cooperation.

In plant physiology, we can find various forms of orientation, such as heliotropism (movement toward the sun) and geotropism (growth toward the ground). Plants also exhibit movement, though on different timescales compared to animals. Plant movement can be tactic, meaning genuine oriented displacements, also found in unicellular organisms, or nastic, involving structural deformations in response to stimuli—these include, for example, thigmonasty, seen in the rapid leaf movements of Mimosa pudica or Dionaea muscipula, and nyctinasty, the closing of flowers at night.

A key molecule in plant signaling is auxin (indole-3-acetic acid), often considered a plant analog to neurotransmitters. Its diffusion is influenced by light and gravity, regulating growth, intercellular organization, and communication. Plants also communicate with each other through airborne chemical signals when they experience damage. For example, jasmonic acid, released in response to injury, can be detected by other plants, prompting them to initiate defensive measures. Some plants even form strategic alliances; when attacked by aphids, certain species release chemicals that attract ladybirds, which prey on the aphids. In Africa, acacias have developed defense mechanisms against herbivores like giraffes—when their leaves are grazed, they send electrical signals to neighboring trees, triggering the production of bitter compounds to deter feeding. The concept of plant intelligence was introduced in Italy by Stefano Mancuso, director of the International Laboratory of Plant Neurobiology at the University of Florence. The 2006 article Plant Neurobiology: An Integrated View of Plant Signalling, of which Mancuso is a co-author, argues that plants, due to their sophisticated interactions with the environment, can indeed be considered intelligent.

This brings us to Artificial Intelligence, a widely debated topic today, after having been the focus of dystopian narratives such as The Matrix and The Terminator, particularly in the late 20th century. The term Artificial Intelligence was officially introduced in 1956 when a group of scholars, including John McCarthy, Marvin Minsky, Claude Shannon, and Herbert Simon, met at Dartmouth to explore the possibility of creating intelligent machines. However, the concept of a machine capable of complex computation and logical reasoning had already been envisioned by the British mathematician Alan Turing (1912–1954) with his idea of a universal machine capable of performing any computational process—the Turing machine. In the 1960s, the idea of an intelligent machine was not only a subject of fascination for mathematicians, psychologists, and science fiction writers but was also beginning to seem feasible with advancements in computer science and cognitive science. Many will recall the emotionally expressive computer HAL 9000 in Stanley Kubrick’s 2001: A Space Odyssey (1968). However, by the 1970s, the limitations of a top-down algorithmic approach to AI were becoming apparent. In a 1973 report to the British government, the scholar James Lighthill highlighted these challenges, signaling a shift in how artificial intelligence development was tackled.

This led to the development of two new approaches. The first involved equipping machines with specific knowledge to enhance their capabilities, as seen in expert systems. A notable example is Deep Blue, the first computer to defeat world chess champion Garry Kasparov in 1996. The second approach focused on connectionist artificial intelligence, based on artificial neural networks. These quickly proved to be more flexible in simulating, at least partially, the central nervous system. More importantly, they enabled the simulation of learning processes, allowing machines to adapt and improve their performance over time. Neural networks are artificial structures capable of learning and performing specific functions based on three hierarchical levels: a perceptual layer that gathers input, a hidden layer that interprets data, and an operational layer that produces output.

These networks can be trained through different methods: supervised learning, where they acquire knowledge from predefined data; unsupervised learning, where they identify patterns or logical structures without explicit guidance; and reinforcement learning, where they improve performance based on feedback from goal achievement. Neural networks simulate the synaptic connections of the brain, often referred to as the connectome. Incidentally. in nature, animals exhibit all three forms of learning, possessing innate heuristics for handling new situations, built-in pattern-recognition abilities, and a pleasure-based reinforcement system that guides behavior. For example, through training sessions such as backpropagation, neural networks can progressively improve their performance. This technique, a form of supervised learning, involves training the network on problems with known correct answers to evaluate its efficiency. The network’s predictions are compared to expected results, and the discrepancy is measured: the smaller the gap, the greater the network’s accuracy. The connection weights are then adjusted to minimize this difference, making fine-tuning more of a supervision process than a simple setup.

Another pivotal moment in the history of artificial intelligence occurred in 2015 with the rise of Deep Learning, a technique that allows machines to learn autonomously and adapt to changes (Y. LeCun et al., 2015). This breakthrough raises a fundamental question: can we now speak of a non-biological intelligence?

To consider the above expressions of purposeful and self-regulating behavior as forms of intelligence is to adopt an expansive definition of the term. However, two narrower interpretations of intelligence exist: the rationalist and the subjectivist conceptions. The first, which has long characterized Western philosophical thought, defines intelligence in logical-conceptual terms. In it, intelligence is confined to conscious acts of doubt, analytical and reflective thinking, where planning and elaboration necessarily involve consciousness. The subjectivist conception, instead, views the mind as a systemic emergence of cognitive faculties. This perspective does not require conscious explication—intelligence can manifest unconsciously—but it is always rooted in internal motivation and thus presupposes subjectivity. In both the rationalist and subjectivist formulations, intelligence is not merely about problem-solving or knowledge expansion but is fundamentally linked to the reflective act of intus legere—asking questions and seeking new possibilities beyond mere performance. The key distinction between the two interpretations lies in the role of consciousness and rational thought. The second, more moderate view—which I prefer—does not require intelligence to be explicitly conscious or rational. This avoids excluding the vast realm of unconscious intellectual processes in humans and downplaying the cognitive abilities of other species. In fact, while consciousness and rational thought are no longer considered exclusive to humans, only certain species can be said to possess both of these functions.

The study of animal intelligence certainly presents challenges that are not typically encountered in the analysis of other biological functions. We readily accept a wide range of perceptual differences among species, yet intelligence remains a particularly sensitive subject, as it is often regarded as the defining trait that sets humans apart. This sensitivity leads to an obsessively anthropocentric approach, where: 1) human intelligence serves as the benchmark against which all other forms of intelligence are measured, placing it at the top of a hierarchy and 2) non-human intelligences are often viewed as mere subsets or approximations of human cognition. This brings us to a fundamental question: is the difference between human intelligence and that of other species a matter of degree (quantitative) or of kind (qualitative)?

In 1983, Howard Gardner, in Frames of Mind, argued that intelligence is not a single, quantifiable entity but rather a combination of multiple independent faculties. According to his theory of multiple intelligences, intelligence emerges from the interplay of different cognitive abilities, producing distinct forms of intelligence: linguistic, logical-mathematical, spatial, social, introspective, bodily-kinesthetic, musical, and naturalistic. To illustrate these distinctions, Gardner associated each type with historical figures—Albert Einstein with logical-mathematical intelligence, Charles Darwin with naturalistic intelligence, Johann Sebastian Bach with musical intelligence, Rudolf Nureyev with bodily-kinesthetic intelligence, and Pablo Picasso with spatial intelligence.

But if this is true for the different propensities within the human population, it is even more evident when considering animal species, each with distinct behavioral and adaptive habits. Intelligence, in fact, is the result of the evolution and specialization of the living—shaped by selective pressures based on the risks, opportunities, and challenges each species has encountered throughout its phylogeny. As such, intelligence can only manifest in a pluralistic sense. In 2008, I published Intelligenze plurime (Plural Intelligences) precisely to emphasize that animal intelligences exhibit far more significant differences than those seen within human beings, as noted by Gardner. These intelligences are not characterized by the overlaps that occur within human multiple intelligences, because they are not the emergent outcome of different compositions of gradients. Instead, they often involve functions that are fundamentally different at their core and are structured organizationally in distinct ways. We could argue that the plurality of animal intelligences, rather than representing a difference in propensities as in Gardner’s multiple intelligences, should be treated similarly to differences in perception. This means that many species-specific forms of intelligence might be incomprehensible or even undetectable to humans, requiring the use of technological tools to study them effectively.

There are countless forms of animal intelligence, and humans cannot be considered the apex or unit of measurement in the exploration of animal cognition. Increasingly, research is revealing the vast intellectual diversity that exists across species, and today, we find ourselves at the beginning of a promising new era of study in this field.

The Study of Animal Intelligence

The debate surrounding animal intelligence often reveals fundamental ambiguities, particularly in relation to the definition of intelligence itself. Some argue that the term may no longer hold epistemic value, and at times, intelligence—and even the concept of reinforcement—can be seen as terms that, rather than helping our understanding, conceal an epistemological difficulty, much like the concept of the epicycle in the Ptolemaic model. While I do not fully embrace these rather extreme conclusions, it is evident that the concept of intelligence is far from straightforward. A concrete example may help illustrate this difficulty more effectively. We all agree that the characteristics of a species have been shaped by natural selection, tailored to the living style and environment of that particular animal population. For this reason, we would never rank species on a performative scale based on factors like walking, the functionality of the endocrine system, foraging methods, or reproductive strategies. So why do we ask what species is the most intelligent, as if intelligence were separate from the adaptive processes that define a species through evolutionary criteria?

The obsession with comparing other animals to human beings prevents us from investigating their intelligence in a purely descriptive way, which is a significant limitation. This bias acts as a veritable hindrance to free research on animal intelligence. In this sense, it is crucial to emancipate ourselves from two opposing forms of epistemological anthropocentrism. First, anthropomorphism, which tends to view the differences between human beings and other species merely in quantitative terms. This perspective suggests that the only difference between human intelligence and animal intelligence is one of degree. Second, anthropocentrism, which recognizes a qualitative difference in the realm of intelligence, but does so by denying other animals any intellectual status.

Following the first argument, the intelligence of other species can only be seen in terms of its approximation to human intelligence: it may have something less, but never more, because other cognitive qualities are excluded from consideration to begin with. Anthropomorphism, particularly non-critical anthropomorphism, might appear to acknowledge animal otherness, but in reality, it strips away the uniqueness of animal intelligence, reducing it to a mere approximation of human cognition. In doing so, it presents other animals as lesser, or “minus habentes“, in comparison to humans.

Let me now turn to the second argument. To deny non-human animals intellectual capacity is to suggest that they lack the ability to have insights into the world, reducing their behavior to mere expressions of blind, mechanistic processes. Those who view animal behavior through the lens of instinct and drive, or who frame learning solely in terms of conditioning, may not explicitly state that non-human animals are mere machines, as Descartes did, but they essentially echo the same sentiment, interpreting animal behavior as a deterministic result of automatic responses. According to this line of thinking, non-human animals would be nothing more than puppets moved by invisible strings—not just poor in world, as Heidegger suggested, but entirely devoid of it. Therefore, I do not think that the anthropocentric claim warrants any further discussion, even in passing.

And yet, methods of investigating the intelligence of other species are often plagued by anthropocentric biases. Common mistakes made when applying analysis methodologies include: (i) the perceptual approach, where tests are often based on orientation and visual choice, reflecting our own preferred sensory modality; (ii) the type of intervention required of the animal, as not all species are inclined toward object manipulation, and even among those that are, there are significant differences in how they manipulate objects; (iii) the way the problem is framed, such as detecting the presence or absence of an obstacle to be removed and an opportunity to be acquired, which is often shaped by the species’ propensity for tool use; (iv) the value assigned to rewards, whether food or social, as well as the temporal framework a species uses to define correlations or cause-and-effect relationships; (v) the relationship with the examiner, particularly the degree of relational ease, which is naturally better between humans than in situations where a human examiner interacts with a heterospecific, as well as referential orientation—how the species uses the examiner as a point of reference to understand what is being asked; and (vi) the type of behaviors that species habitually engage in and the consistency of those behaviors with the actions required by the examiner, as each activity must align with the species’ innate propensities.

The concept of plural intelligences that I have formulated aims to overcome both arguments, proposing instead a plurality of intellectual diversity within a continuum, much like other functions derived from phylogeny. This perspective takes into account two key aspects: first, the diversity of cognitive qualities that combine to form the intelligence of a species; and second, the diversity of the animal’s system for processing incoming data and generating responses—whether this system is internal to the organism or extended beyond it, whether it is collective or group-based, and whether it is centralized, diffuse, or ganglionic in nature. Moreover, It is essential to distinguish between performances that appear to be the result of local, individual-specific processing and those that seem to arise from innate instructions embedded in the species. Yet, we would be mistaken to oppose these two terms, as every innate expressive pattern always requires an adaptive intervention—what we might call local processing—on the part of the individual (adaptive intelligence).

For instance, consider the architectural design of a hornet’s nest, which, despite its complexity, is identical across all communities of that species. When a female queen decides where to build the nest, she must be able to evaluate various options, even though she is guided by innate coordinates. Ultimately, she must make a decision based on an evaluative act. Moreover, when the queen proceeds to build the nest, she will undoubtedly rely on a set of innate design behaviors and correction systems based on sensory feedback interactions. By utilizing her local processing capabilities, she can adapt the nest to the specific conditions of the chosen site, taking into account its constraints and opportunities. This adaptation process implies a form of supervision in the task; therefore, the relationship between innate instruction and local processing is not inversely but directly proportional, as each element supports and enhances the effectiveness of the other.

That being said, the assessment of animal cognition is not easy and requires careful planning of observation and testing methodologies to avoid certain biases in judgment. One such bias is perspective bias, which occurs when we prioritize our own point of view, assuming that it should be the same for other species. This bias often leads to the assumption that other animals must pass the same tests routinely applied to humans, typically in the same testing environment and using the same approach. This perspective diminishes the value of differences and specificities inherent to each species; as a result, there are numerous distortions present in the study of animal intelligence. These not only stem from the risks of anthropomorphism and anthropocentrism but also from an inability to accurately assess intelligence as something beyond human characteristics. As mentioned, the setup of the test itself introduces several anthropocentric pitfalls, such as how the setting is organized, the problems posed to the animal, and the performance expected of it. Vinciane Despret, in her essay What Would Animals Say If We Asked the Right Questions? (2016), highlights these issues, pointing to the epistemological obstacles that hinder the accurate study of animal intelligence.

Intelligence and Anthropocentrism

Intelligence may seem to defy the constraints of specialization, which by its very nature cannot aspire to absolutes or neutrality. This is particularly true because intelligence is an animal’s ability to confront critical issues, solve problems, adapt behavior, and seize opportunities. However, these abilities are always specific to an animal’s lifestyle and the particular environment it inhabits. Intelligence is, therefore, a trait like any other, subject to selective pressures that do not reward performance in absolute terms, but rather that which is most productive within the ecological context in which the species exists. The fitness of a species’ intelligence is determined by how well it correlates with the challenges and opportunities faced by that species. What is rewarded is not performance in absolute terms—provided intelligence could ever be defined as absolute—but the cognitive specialization that has been achieved. Certain problem-solving, creative, investigative, and adaptive abilities may be entirely irrelevant for one species, while being of absolute necessity for another. In analyzing the intelligence of other species, we often make two fundamental errors: first, we tend to use the human being as a unit of measurement or a benchmark for comparison, viewing our species as an absolute and unspecialized entity; second, we mistakenly regard intelligence as a function that transcends phylogeny, considering it a trait whose value is unrelated to the species’ other biological characteristics, and one that can be assessed through a performance scale.

In case 1, we are dealing with epistemological anthropocentrism, whereby human beings claim to be a neutral focal point. This perspective implies, in simpler terms, two things: first, that human intelligence is assumed to subsume all potential forms of intelligence, and second, that human cognition is considered free of biases that might distort the way we interpret intelligence. In practice, this translates to the idea that humans are both the standard for measuring intelligence and its highest expression, defining its essential qualities while assessing other species’ intelligence only by approximation to their own. These two assumptions present an epistemological stumbling block because they prevent us from assessing the cognitive performance of other species in a purely descriptive way. Instead, this framework forces us to view intelligence through comparison with humans. Consequently, it follows that: first, the most intelligent animals will be those closest to us from a phylogenetic standpoint, and second, any performance that differs from human performance will automatically be disregarded or deemed less significant.

This leads us to a fundamental question, which has been raised by several authors: namely, whether it is possible to overcome the anthropocentric perspective. I believe this question can be clarified in a positive sense by two key findings. The first concerns the scientific method itself, which often employs a counterintuitive approach. Much of scientific theory is based on overcoming anthropocentric explanations, as demonstrated by revolutionary ideas such as the Copernican model of the universe or Darwinian evolution. The second finding comes from ethological analyses of other areas, such as sensory perception, which have revealed how perception can vary significantly across species. So while epistemological anthropocentrism is not entirely surmountable, it can certainly be amended, criticized, or confined to a domain of validity. Unfortunately, with the exception of a small group of ethologists and neuroscientists, the majority of the humanistic world still regards human intelligence as the pinnacle and touchstone of intelligence, if not the only valid form. As a result, the intelligence of other species is often evaluated based on how closely it approximates the human. This approach fails to recognize intelligence as a systemic function that, like other morpho-functional traits, is the result of specialization—specifically cognitive specialization. Thus, it inevitably introduces an elaborative bias into the data it addresses.

If case 1 is epistemological anthropocentrism, case 2 is intellectual transcendence, where intelligence is not seen as an adaptive function designed to oversee behavior and rooted in innate characteristics but as something that transcends adaptation and can be valued independently of the evolutionary lens of specialization. When we analyze other functions like perception or nutrition, we do not fall into the trap of seeking an absolute value—the best way to perceive or feed—but we understand how each function is related to phylogenetic adaptation. No function is absolute, as phylogeny does not pursue the best performance in a general sense, but the most adaptive one—one that best fits the ecological niche of a particular species. A function is adaptive not because it is the absolute best, but because it correlates with the selective pressures faced by the species in its phylogenesis. Intelligence, however—perhaps because it is seen as a point of pride for our species, something to be defended from the relativism introduced by Darwinism—is often detached from the principle of phylogenetic specialization. It is treated as a generic problem-solving function, divorced from the specific goals of a particular species, as if problems were ever absolute, unrelated to the unique, constitutionally specific interests of each taxon.

Upon closer inspection, these two epistemological errors converge in the tendency to treat intelligence as a faculty that should not be subjected to adaptive evaluation, sometimes going so far as to claim that intelligence is inversely proportional to phylogenetic endowments. In practice, it is asserted that the more pronounced the instincts—the innate ability to perform a behavioral task—the smaller the space for intelligence. This view overlooks the fact that cognitive a-priori frameworks, by definition innate, are the foundation upon which experience and knowledge are built. The tendency to decouple intelligence from other behavioral functions leads to the mistaken belief that it is not rooted in the particular needs of a species. Why, for example, would a species shaped by the autonomy of its somatic endowments, like cats, need to use tools? Why should a species that evolved as a solitary animal be endowed with social intelligence? Moreover, there is a tendency to decouple intelligence from other components of the psychic condition—such as emotions and motivations. This view regards intelligence as an autonomous and self-founded entity, explicable iuxta propria principia. Today, this conceptualization often associates intelligence with problem-solving abilities and computational skills, partly due to the information technology revolution that characterizes our age.

There is no doubt that giving a singular definition to the different forms of intelligence or finding a concept that subsumes them is a difficult task, for a number of reasons. First, when we look at the intelligence of different animals, we can detect marked differences in specific performances, making it more accurate to speak of intelligence in the plural, especially when comparing individuals of different species. For instance, there are species highly endowed with technical intelligence, such as primates—this is no coincidence, as humans belong to this order. Other species exhibit exceptional motor organization skills, placing them alongside kinesthetic intelligence, such as most felines. Some species, like cetaceans, demonstrate marked communication skills and creativity in constructing complex and referred semiotics. Instead, social intelligence is characteristic of species like wolves, African wild dogs, and hyenas.

In conclusion, it is impossible to create a quantitative index of animal intelligence that assigns a place of honor to the most intelligent species. Intelligence, like all other functions, is a specialization that has evolved during phylogeny to meet specific, qualitative needs. Therefore, there is no such thing as a “most intelligent” species.

When can a function be said to be intelligent?

If we consider the ability to overcome limitations or solve problems, there is no doubt that there are various forms of intelligence in nature. The intelligence exhibited by an animal through its psychological condition and behavior is just one of the many expressions of nature that appear intelligent to us, as they are oriented, modulated, self-regulated, correlated, and productive. We could even say that the biosphere itself, with its mechanisms of systemic regulation and mutual dependence, manifests an intelligence of its own. After all, the process of adaptive specialization can also be considered a form of cognition—phylogenetic intelligence—that endows the animal with a set of ready-made solutions. This intelligence results from the evolutionary mechanism, which is based on the production of variants to be screened by natural selection for reproductive capacity (fitness).

When we observe a spider constructing a web with perfect radial symmetry or a dog bred for hunting enacting perfected hunting patterns without being taught, we cannot help but admire the form of intelligence they display. Through the joint work of three factors—variant production, natural selection, and character inheritance—over the course of biological time, different species achieve extraordinary performative correlation. So much so that we could think of species as a particular way through which life has specifically solved certain problems. All phylogenetic characters reveal incredible abilities to address the unique challenges posed by a species’ lifestyle and environment. But can we say that this species-specific intelligence, fine-tuned by natural selection, is attributed to an intellectual effort made by the individual? Yes and no. No, because that behavior is already innately present in the individual, so the animal does not need to solve the problem from scratch—it already has the solution within. Yes, because the innate not only provides the individual with preformed solutions, but also with the foundational material that sometimes enables new solutions to emerge, and in any case, prompts the individual to adapt instinctive behavior to the specific situation.

The matter becomes more complex when we consider the genetic characteristics of a species not merely as an instruction manual for the phenotypic assembly of the organism, but as an entity that is sensitive to external information. The genome, in fact, contains multiple translation possibilities (polyphenism), and the choice of these depends on the ecological situation in which the developing organism finds itself. At this point, we must speak of genetic intelligence—the ability of the gene to adapt to situations, allowing the development of different morpho-functions based on the organism’s developmental context. Alongside this, and in competition with it, we observe an epigenetic intelligence, arising from the residual organic portions of the fertilized egg cell (zygote)—in the form of proteins, methylations, and RNAs—that also interact with the genome and influence the phenotypic translation process, silencing or emphasizing certain genes. Once again, we see how the organic constitution of living beings is intelligent in the sense that it is intrinsically capable, by inheritance, of providing correlated responses to the challenges posed by the environment.

Another form of species-specific intelligence involves the ability to control the environmental situation, rather than being completely at the mercy of external fluctuations. This is known as niche intelligence. It is well established that living organisms do not merely occupy an empty niche; they actively reorganize their surroundings (niche construction) through the reconfiguration of ecological values, activation of symbioses, and the creation of elements designed to buffer fluctuations in environmental gradients. Moreover, the niche is an integral part of a species’ inheritance. In relation to the niche, we can also refer to the concept of extended intelligence, proposed by A. Clark and D. Chalmers in their 1998 article The Extended Mind. According to this theory, cognitive processes result not only from what is inside the organism but also from the surrounding environment and the tools the individual uses. This idea is known as active externalism, which emphasizes the role of external devices in guiding intellectual capacity. For example, we might say that a spider not only perceives the world through its internal nervous system but also through the web it constructs. Every external tool, in fact, alters how an organism cognitively interprets its relationship with the surrounding reality. This concept aligns with the theory of affordances, introduced by the American psychologist James Gibson in 1966, which suggests that the form of external objects inherently calls for certain actions, essentially offering a cognitive “invitation.”

Another form of intelligence involves the ability of living systems to learn through interaction with the environment, due to the inherent properties of the biological substrate, such as organizational flexibility, differential growth of parts, internal selection processes, recycling of functions, and the construction of new connections.

To conclude this discussion, we can identify several characteristics that suggest intelligent behavior in a system. These include: (i) functional relatedness, or the ability to adjust a function depending on the context; (ii) goal orientation, the capacity to direct actions toward an outcome, even without an intrinsic purpose embedded within the system itself; (iii) the ability to modify itself through experience or as a result of actions; (iv) the ability to overcome challenges, addressing critical issues and problems to achieve goals; (v) self-regulation, or the ability to modulate a function based on the outcomes achieved; and (vi) the capacity for multiple levels or functional states, rather than adhering to a single operational plan. These properties are commonly observed in nature, not only within behavioral domains but also in basic physiological processes. Furthermore, today we have machines that, in varying capacities, perform functions demonstrating these properties, including the more complex behaviors exhibited by artificial intelligence.

Clearly, there are various forms of intelligence, and any definition will always be partial, so I have no claim to be exhaustive. What I’ll try to do in what follows is examine the differences between artificial and animal intelligence.

The “something more” of animal intelligence

If we can identify basic mechanisms that indicate a process is carried out intelligently, we must also recognize that when we speak of animal intelligence expressed in behavior, we are referring to something else. While not necessarily separate from these functions, in fact, it is endowed with something more, so to speak. This surplus, in my view, is not about awareness, although it is evident that, on many occasions, awareness may be present. Instead, it pertains to the subjective condition of the animal. In analyzing the five levels of subjectivity, I have particularly focused on the second level, the affective dimension, which enables a subject to recognize itself as having inherent interests. These interests relate to the fact that an animal is a body: i) endowed with specific needs that demand fulfillment; ii) regulated by a hedonic system that leads it to avoid pain and suffering while seeking pleasure; and iii) influenced by emotions and motivations. Physical needs and the sense of pleasure or suffering give rise to psychological conditions that drive the animal toward specific goals. The same can be said for emotional and motivational dispositions.

Affectivity gives the animal an intrinsic purpose, meaning it is oriented toward something and motivated to achieve specific ends or goals. The animal’s intrinsic purpose, however, expressed through psychological states and behavior, should not be confused with other forms of teleology, such as extrinsic teleology. The latter interprets the evolutionary process as being directed toward an ultimate purpose—a principle I believe should be rejected. Intrinsic purpose is derived from: (i) elective orientation, which enables the individual to make choices and always be directed toward something to be achieved; (ii) agency, the ability to oversee events, navigate through the world, avoid risks, and seize opportunities; and (iii) self-referential sensitivity, which drives the individual to protect their own integrity and defend their interests. These qualities are always expressed circumstantially, meaning they are shaped by the specific conditions the individual faces. Intrinsic purpose, therefore, is an attitude toward the world that grants the subject ownership of their presence—through seeking, problematizing, choosing, and deciding. Intrinsic purpose cannot be reduced to teleonomy because of its deeply subjective and context-dependent nature, which has nothing to do with a mere mechanistic process. Affectivity brings forth a desiring condition, one that is open to constructing goals, and from this condition emerges the relationship between the individual and the problem at hand.

Can we then consider a problem in absolute terms? From my point of view, this perspective is incorrect because it is always the individual who poses a problem in the particular circumstance where they intersect with the world, with its possibilities and limitations. Therefore, we can say that there are no absolute problems, because a problem always emerges based on the objectives the animal sets in a context-specific way, engaging directly with what is offered in the here and now of the situation. On the other hand, it is clear that the goal translates into a problem because it goes beyond what the external world provides. This means that intelligence, like all other operational functions, serves the affective components—physical needs, pleasure, emotions, motivations—and can never be separated from the subjective needs that drive the animal to project itself beyond what is within its grasp. From this tension arises the need to identify solutions. Therefore, it is impossible to think of intelligence as independent of phylogeny.

We often focus exclusively on solution intelligence, the ability to solve problems, and give little attention to problem intelligence, the ability to pose problems. However, there is no doubt that in animals, the former is strictly dependent on the latter. This is not the case with artificial intelligence, as it is the human operator who imparts problem intelligence to the machine, driven by their own motivations. In other words: i) affectivity brings out a purpose, the goals the individual sets; ii) the presence of goals in relation to the world activates problem intelligence in the individual; iii) the problems the individual sets activate the search for solutions. Without these elements, which stem from the animal’s affective condition—encompassing need, pleasure, and disposition—there can be no problematic attitude, and consequently, no effort to find solutions. If we agree on this, it becomes evident that since the affective condition varies across species, each type of animal will have different problematic attitudes, meaning it will pose different problems.

We are accustomed to thinking of intelligence in terms of input processing, without reflecting on the problem attitudes of different species. However, I believe this is a partial reading of intelligence, and the exclusively computational or solution-oriented view of intelligence raises several concerns. If we consider alternative modes of interpretation, we find, for example: (i) the concept of intus legere, that is, delving beyond appearances or habit to seek new possibilities; (ii) the philosophical principle of asking questions, or doubting, which subjects reality to stress tests and investigations; (iii) the idea of behavioral superintendence, the ability to adjust an innate organizational plan—such as a bird’s nesting behavior—adapting it to the specific conditions found in the environment; (iv) the principle of creativity, which some authors, like Bergson, argue is the driving force of the evolutionary process; (v) the will to affirmation, the principle of expansion that drives individuals to transcend their own conditions and, thus, to imagine; and (vi) the intentional view of intelligence, seen as a copulative function that situates it in reference to something to be negotiated in relationship with the world.

It is often assumed that intelligence is a self-founded faculty but in reality, all cognitive expression is driven by the affective state of the individual. The individual, in fact, strives to understand and learn because it is motivated by what I have called intrinsic purpose. Each species has a unique condition of intrinsic purpose, as a response to external conditions and a search for present opportunities. Therefore, the problem intelligence of one species is not comparable to that of another. For instance, if we compare dogs with cats, we observe that in dogs, the collaborative system prevails, where they seek to develop group actions and strategies based on concerted efforts and role differentiation; in cats, instead, a solo approach is dominant, where they seek solutions suited to themselves, without involving or coordinating with others. The motivational profile of a species, therefore, indicates its intentional motives—what behaviors it engages in, what stimuli interest it, what type of feasibility it seeks, and what elements of recall and orientation prompt elaborative action. There is no such thing as knowing for its own sake because cognitive processes are driven by the needs and interests specific to different species. For this reason, when evaluating results in different learning or problem-solving tasks, one cannot only consider the cognitive characteristics of a species but must also account for its perceptual, operational, and motivational aspects.

If we aim to identify the key factors behind the intellectual differences among species, we will find that this diversity is far greater than the variation observed within the human species. In the animal world, there exists a true intellectual plurality, comparable to the diversity found in sensory perception. This should encourage caution when attempting to assess the intelligence of a given species. Indeed, even within mammals, as we move beyond primates, we encounter sensory features that are entirely foreign to human experience, such as echolocation. How, then, can we begin to comprehend the sensory apparatus of fish, insects, or mollusks? We must apply the same reasoning when encountering alien intelligences, such as those of an octopus or a praying mantis. To make another example, the swarming intelligence of hymenoptera differs from ours not only in performance capabilities but also in the very nature of the cognitive process itself. This perspective compels us to consider intellectual diversity among species as something far more profound than mere specialization. Instead, it should be understood in terms of: 1) The package of cognitive qualities, the specific processing specializations that have evolved in a given species, and 2) the data processing system, how the species’ neurobiological structures and, more broadly, its psychoneuroendocrine-immune (PNEI) system are configured.

Let us start with point 1. There is no doubt that intelligence varies significantly between species. What differs primarily is the package of cognitive qualities that a species has developed, including:

1. Operational memory – highly developed in felids.
2. Spatial memory – crucial for species that store food in multiple locations, such as the Eurasian nutcracker.
3. Episodic memory – prominent in cetaceans and great apes.
4. Ability to distinguish and categorize – well-documented in parrots.
5. Assessment of object spatial orientation – examined in pigeons.
6. Tendency to manipulate and use tools – characteristic of primates.
7. Kinesthetic performance – highly refined in cats.
8. Complex communication and dialect construction – found in primates and cetaceans.
9. Capacity for biographical memory and individual recognition – present in various species, including Polistes wasps.
10. Attribution of qualities to mates, such as assigning names – observed in elephants.
11. Symbolic attribution and inference abilities – documented in chimpanzees.
12. Development of traditions and cultural differences within the same species – seen in killer whales and chimpanzees.

Now, let us consider point 2, which concerns the diversity of processing systems. The anatomical-functional constitution of a species and its relationship to the world determine different modes of systemic processing. While vertebrate intelligence is typically centralized, with most cognitive functions coordinated by the encephalon, other species exhibit distinct processing architectures, including:

1. Extended intelligence – characteristic of species whose cognition is deeply intertwined with their niche and external structures, such as the spider with its web or humans with their technological environment (the technosphere).
2. Emotional intelligence – which enables a species to assess situations through emotional configurations, providing an affective marking of circumstances.
3. Collective/swarm intelligence – found in eusocial species like ants and bees but also observed in flocking birds and schooling fish.
4. Diffuse intelligence – present in animals like octopuses, where neurons are distributed throughout the body, enabling decentralized processing.
5. Ganglionic intelligence – typical of insects, where different body parts are governed by relatively independent neural clusters, each managing specific functions.
6. Visceral intelligence – which arises from neural activity in visceral centers, such as the gut-brain axis.

Recognizing these multiple forms of intelligence challenges anthropocentric biases and underscores the need for a broader, species-specific approach to understanding cognition.

The Theory of Plural Intelligence

The intelligence of a species, like its innate competences, is the result of its progressive adaptation to its niche—the interplay between the species’ environmental modifications and responses to them. Each species, alongside its morphological and behavioral traits, thus possesses a distinct intelligence, meaning a characteristic way of approaching reality, identifying problems, and adapting its skills to specific situations. We can think of intelligence as a species’ mode of reasoning—its ability to reflect on the hidden aspects of phenomena and possibilities to explore. This reasoning process depends on the coordinated activity of various functional components, such as working memory, dominant logical structures, dispositional tendencies, and attentional focus. At its core, intelligence is the ability to go beyond the obvious and already-experienced, to uncover latent possibilities, and to work creatively with contingent data rather than simply applying pre-established skills in a rigid manner. Intelligence challenges situations, digs beneath appearances, rummages through the meshes of hidden possibilities, and poses problems to the world.

For example, an animal with high social aptitude, such as a dog, is not only equipped with strong communication, mediation, and engagement skills—traits that facilitate relationships—but also perceives social interactions as complex problems to be navigated and interpreted in order to generate new opportunities. Solitary animals—such as most felids, mustelids, and certain canids like the fox—tend to exhibit a predominantly solutive intelligence, meaning their cognition is geared toward analyzing the specific requirements of problems and maintaining a strong curiosity for situational particularities. This intelligence appears highly innovative because it relies on personal ingenuity to develop ad hoc solutions. In contrast, social animals perceive problems not as individual concerns but as challenges that affect the entire group. Their approach is less about direct confrontation and more about constructing an appropriate social strategy—a cooperative game plan—tailored to the situation at hand. Some species are particularly predisposed to using external objects as extensions of their bodies, favoring tool use and specific motor skills over full-body movement. Others, by contrast, excel in athletic abilities, demonstrating high levels of psychomotor coordination and kinesthetic creativity. It would be a mistake to consider solutive intelligence superior to social intelligence, just as it would be misguided to assume that an animal using tools is inherently more intelligent than one with exceptional physical coordination.

Just as bodily structures exhibit internal coherence—for example, the adaptation of the entire musculoskeletal system for human bipedalism or the integrated body structure of birds for flight—so too there is a close relationship between a species’ lifestyle, ecological niche, innate abilities, and cognitive skill. Intelligence is a complex interplay of fundamental cognitive functions—such as problem-solving, memory, foresight, simulation, spatial orientation, reflection, and self-awareness—which manifest differently across species, shaping distinct modes of intelligence that cannot be ranked on a hierarchical level.

Plural Intelligence Table

Observing animal behavior, I identified (2008) these broad intellectual categories, each characteristic of certain species.

The study of animal intelligence is central to this broader reflection on creativity, as it reveals that an individual’s behavior cannot simply be reduced to innate endowments. Rather, it is precisely the existence of such endowments that necessitates an individual’s ability to apply, adapt, and innovate. In other words, behavior is always shaped in the present moment through complex reasoning, where innate capacities function as tools that must be continuously adjusted, refined, and corrected.

However, research on intelligence also compels us to view the animal world as an existential pluriverse. Just as there is an immersive Umwelt, we must also recognize a cognitive Umwelt, demonstrating that every animal expression—though bound by a range of possibilities—is, in its own way, an act of world-creation. At times, this perimeter of possibilities is stretched, as individuals or populations embark on new existential trajectories, driven precisely by the creative power of intelligence. Speciation, therefore, is not merely the product of genetic drift but also the outcome of transformations in species recognition systems—that is, in the ways individuals establish belonging, including through aesthetic experience. Intelligence never treats anything as fixed or final. The time has come to abandon our anthropocentric biases and appreciate intelligence, in its plural manifestations, for what it is.

Featured image:

Karin Andersen, Listener, 2003, cm20x30, fotografia digitale su alluminio.