“ASSESSING A MULTILEVEL CAUSALITY MODEL IN THE OF ”

MSc in and Cognitive

Behavioral track

By: Vanessa Del Pozo Sánchez

Student no: 11104066

First assessor: mw. dr. Federica Russo Second assessor: dhr. dr. Hein van den Berg

Date: 28/June/2017

Abstract

Cognition is the by which an organism is able to acquire, process, and retain information through senses and experiences. To this day, there is no agreement regarding its evolutionary explanation. The literature describes a wide range of methods in order to understand components such as motor , thought, , , , and language, among others. This is done with the aim of complementing the origin of the evolutionary processes of cognition. However, none of these approaches has considered the evolution of cognition as a result of a network of complex interactions at different levels of organization. In this thesis, we introduce a multilevel causality model for the understanding of the adaptationist idea of cognition. The model is built by integrating three evolutionary processes: ontogeny, phylogeny, and Evo-devo. The model allowed us to fill the failures that evolutionary processes presented. Thus, we can conclude that with more detailed studies of multilevel causality in the biological systems of cognition, we can develop complete explanations of evolutionary mechanisms that occur at certain level, and observe their consequences at other levels.

1 Index

General introduction 3 1. Historicity of cognition and Dennett’s adaptationist approach of cognition 4 1.1 History of cognition from Lamarck onwards 5 1.2 Cognition: an adaptationist property according to 10 1.3.1 Evolution of simple entities by Dennett 11 1.3.2 From simple replicators to the 11 1.3.3 Phenotypic variation and the understanding of cognition 13 2. Ontogeny and the “ of neuronal group selection” in the understanding of cognition. 15 2.1 Introduction 15 2.2 Edelman and the variation and selection within neural populations 15 2.2.1 The three main tenets of the theory of neuronal group selection. 16 2.2.2 Degeneracy and value 18 2.3 Edelman’s theory applied to the 19 3. Evo-Devo: an extension to cognition 20 3.1 Introduction 20 3.2 Evolutionary developmental theory 21 3.2.1 The neural system from an Evo-Devo perspective. 22 3.2.1.1 Types of constraints 23 3.3 Evo-devo and cognition 24 4. Multilevel causality models and cognition 26 4.1 Introduction 26 4.2 The understanding of adaptation 26 4.3 Difficulties in the explanation of cognition by the adaptationist approach. 27 4.3.1 Difficulties in Dennett’s approach and ontogeny 27 4.3.2 Analysis of Edelman and Dennett’s : differences and similarities. 30 4.4 Multilevel causality 31 4,4,1 Types of multilevel causality 31 4.4.1.1 Bottom-up causality for the understanding cognition. 32 4.4.1.2 Top-Down causality and cognition 34 4.5 Integration of types of multilevel causality model to understand cognition 36

Conclusions 38

References 40

2 Introduction

Evolution refers to the process by which species respond and adapt to the environment as a result of maximizing their fitness. Changes occur at all levels -from protein synthesis to behavior- in order to predict and even manipulate environmental regularities. One category of such mechanisms of response and adaptation is cognition: the process by which an organism is capable to acquire, process, and retain information through senses and experiences. Therefore, cognitive capacities in animals -including the being- can be studied as phenotypic traits within .

The general objective of this work is both to show how cognition develops and to address those of its elements that allow an adaptive explanation. We focus on evolutionary processes through ontogeny, phylogeny, and evolutionary , all of which deal with the development of cognition in animals with a central nervous system. Then, by combining the study of these evolutionary processes, we can apply the multilevel causality model as a tool to understand the evolution of cognition in the adaptational approach.

This work will be structured in the following manner:

- To identify the epistemically significant features of cognition, it is necessary to know the philosophical and historical approaches by which it has been studied. In the Chapter 1, we present how the understanding of cognition has developed trough , and then we include an analysis of the main ideas of one of the ultra-Darwinian authors in adaptation of cognition: Daniel Dennett.

- In Chapter 2, we describe an evolutionary take on cognition from the ontogenetic perspective. We analyze the theoretical arguments of neural given by the neurobiologist , the essence of whose argument is based on an analogy of evolution by Darwinian natural selection at the cellular level. Edelman argues that a selection process is carried out at the level of neural groups in the nervous system of some . With this, he explains the emergence of such cognitive phenomena as perception, memory, and consciousness.

3 - In Chapter 3, we review the explanations from developmental evolutionary biology (Evo-Devo) regarding the origin of the variations for which natural selection applies. In order to understand the weight those variations have in an adaptive explanation of cognition, we also examine the role of natural selection itself in the arguments of the Evo-Devo.

- In Chapter 4, we develop a multilevel causality model. To do this, it is essential to understand what a multilevel causality model is and how does it work. After that, we construct a model of cognition that integrates ontogeny, phylogeny, and evolutionary developmental biology. Our motivation stems from the idea that biological systems are organized hierarchically in levels, which range from molecules to ecosystems. Lower levels limit the higher ones, but these in turn also influence the first in a reciprocal causal dependence. Finally, in the last section, we present the conclusions of our work.

4 1. Historicity of cognition and Dennett’s adaptationist approach of cognition

1.1 History of cognition from Lamarck onwards

Even though there have been many attempts to categorize living beings, from Aristotle to Cuvier, it was not until the early nineteenth century that the French naturalist Jean B. Lamarck –in his work Philosophie Zoologique (1809)– created a classification of the animal based on the degree of intelligence that each animal presents. As such, Lamarck can be seen as the first one to study cognition with an evolutionary approach (Atran 1993).

The classification proposed by Lamarck was based on functional correlations of brain structure, and it rendered three large groups: apathetic, sensitive, and intelligent. Within the apathetic group, we could find cnidarians, , and worms. Crustacean mollusks, insects, arachnids, echinoderms, and myriapods were part of the sensitive group. Finally, the intelligent class was made up of all .

Based on his classification, Lamarck divided the animals into evolutive stages. He devised a new system –the system of “perfection”–, where the simplest existing animals rose progressively to the most complex or “perfect” animals. Just as in the “Scala naturae” of Aristotle, the highest level in this stepwise process was occupied by . Only at the highest level did the organisms display psychological functions such as memory, judgment, attention, and thinking (Papini 2009).

In the 1870’s, , another revolutionary figure of the natural world, published two books that referred to the evolution of the . His work “The Descent of Man and Selection in Relation to Sex” (1871) established that mental characteristics –including moral and social instincts in humans– are inherited in the same manner as physical characteristics, namely by variation and Natural Selection (NS). In the other book, entitled “The Expression of in Man and Animal” (1872), Darwin brought into discussion the concept of emotion. He claimed that are just like any other characteristic, so that they too undergo adaptations and evolve. He compared facial expressions of some primates against humans’ facial expressions, finding out that facial expressions were sometimes caused by desires and sensations unleashed by the nervous system.

5 Lamarck and Darwin were revolutionary naturalists that both gave way to a new paradigm that featured ideas we now associate with cognition. In fact, Alfred Giard, an important zoologist who published his findings during the 1880’s, considered that Lamarckism and Darwinism were actually complementary theories, and he supported this idea with studies on evolutive convergence. For example, he put forward a classification of organisms based on their behavior in their natural environment, and did this with a Lamarckian approach (Peláez del Hierro et al., 2002).

Carrying on with our chronological review, we must mention Douglas Spalding, who is regarded as one of the founders of ethology and who published his first work, entitled “On instinct”, in 1872. His overall studies brought him to the conclusion that instincts are a guide for learning and for inherited capacities (Thorpe, 1979; Gray, 1967). Similarly, George Romanes, an evolutionary biologist and psychologist who set the foundations of comparative psychology, proposed general principles for the evolution of the mind based on psychological capacities rather than physical relationships between animals. He presented them in his book “Animal Intelligence” (1882).

Karl Lashley, a behaviorist from Virginia, carried out studies on intelligence, behavior, and the neuronal basis of certain cognitive processes as of the 1920’s. He focused on brain and psychology, trying both to find the locus of specific memory traces and to describe the behavior of the mind with mathematical and physical concepts. His contributions to the study of cognition were based on the different types of tests that he ran throughout his life. He sought to understand the inconsistencies between different types of learning tests for a variety of different animals, which spanned from rats to monkeys (Lashley, 1929; Lashley, 1950; Lashley 1951).

In the middle of the 20th century, authors such as Niko Tinbergen and Konrad Lorenz, Nobel Prize winners for their work in organization and elicitation of individual and social behavior, rewrote the concepts of ethology. Tinbergen introduced four basic questions in order to characterize an evolutionary approach of behavior (Bateson and Laland, 2013): I) What is the objective of a given behavior? II) How did it develop during the lifetime of the individual? III) How did it evolve over the history of the species? and IV) How does it work? Lorenz, in turn, studied behavioral phylogenies –for which Natural Selection also takes place–, as being guidelines for animals’ instincts (Papini, 2009). Both authors worked under the scope of Darwinism, and they brought it about that evolution of behavior was treated as part of the process of NS. Following this approach, William Hamilton, a theoretical evolutionary biologist considered as one of the biggest influences of the 20th century, conceived a mathematical

6 model of that incorporated both the “coefficient of relationship” –a concept defined by Wright– and the maximizing property of “Darwinian fitness”. The combination of these theories allowed him to model the link between the fitness of species and the evolution of behavior, based on the interactions among such species (Hamilton, 1964). Shortly after, and as a follow-up of his previous work, Hamilton published an article that used neo-Darwinian principles to explain the behavior of a society as based on paternal care (Hamilton, 1964).

In 1975, Edward Wilson, an evolutionary and socio-biologist, defined the systematic study of social behavior by the integration of three different factors: population’s genetics, evolutive , and demography. With this, Wilson tried to extend the focus to other forces that could be guiding behavior as well, and did so with a neo-Darwinian approach.

In 1971, Daniel Dennett, one of the most outstanding philosophers of due to his work in the cognitive sciences, proposed that animals have beliefs. He considered beliefs as “cognitive states that suffice to account for the perceptualocomotory prowess of animals” (Dennett, 1971). With such ideas, one would be able to predict behavior by adopting an intentional stance1.

In his book “The Architecture of Cognition” (1983), the natural philosopher John Anderson tried to explain what is referred to as the modular approach to cognition. He considered that cognition is a process built by quasi-independent modules that become associated with each other in order to construct a higher function. In this multi-module and multi-level process, the degree of complexity for said construction is determined by the size of the brain, a hypothesis commonly known as “encephalization hypothesis”. His theory established that there is an allometric relationship between body mass and brain in all mammals. The hypothesis also states that mammals with brain size bigger than “normal”, where the term “normal” depends on their body mass ratio, also have enhanced cognitive abilities (Boddy et al., 2012).

Patricia Churchland, a philosopher contemporary of Anderson who has contributed to the fields of neurophilosophy and of mind, studied the relation between mind and brain, focusing on the role of neuroscience in within a philosophical context (Churchland, 1984; Churchland, 1989). She argued that in order to have an understanding

1 According to Dennett, there are three different strategies that we might use when confronted with objects or systems: the physical, the designs and the intentional stance. We use each of these strategies to predict and thereby to explain the behavior of the entity in question. Particularly, when he refers to an intentional state, he refers to the mental states such as beliefs and desires which have the property of “aboutness,”that is, they are about or directed at, objects or states of affair in the world (Jones, 2013).

7 of the mind, we first needed to understand the brain. For her, consciousness does not exist; it is just an epiphenomenon of a cerebral function and should be considered only as term that humans have developed with aims of understanding such a function. According to her, such a term will eventually disappear from science.

None of the authors mentioned above studied cognition from a truly biological standpoint. For example, Daniel Dennett used to refer to “cognition” as the process of manipulation of information. As opposed to these takes, then, Fiddick and Barrett (2001) aimed to give a suitable explanation of cognition as ensuing from the concept of Natural Selection. They considered that adaptive cognition should be studied by taking into account a) natural history, b) evolutionary changes, and c) an ecological, functionalist perspective. Thus, they advocated for an interdisciplinary study that included different methods, laboratory studies, phylogenetic and comparative approaches, developmental studies, and neurophysiological dissociations.

Supporting Fiddick’s and Barret’s ideas, Daniel Dennett himself suggested to study cognition from a different angle, since up until the 20th such a study had been confined to an ideology and methodology based on old-fashioned experimental psychology. In turn, he proposed that in order to complement the biological studies of cognition, science should also tackle both the concept of “the mind” and the mental states commonly known as desires and beliefs. With this wide-reaching approach, it would be possible to develop models, theories, and explanations that could prove useful in the understanding of rational agents. This new approach is nowadays known as “cognitivism”. Cognitive authors do not believe in the existence of a soul or ego that rules someone’s behavior. They rather interpret the mind and mental states as ensuing purely from the physical qualities of the brain, and have concluded that there should be no distinction between mind and brain.

For authors like Merlin Donald, a Canadian psychologist, cognitive neuroscientist, and neuroanthropologist, cognition is “the mediator between brain and culture”. In this case, human cognition is seen as having emerged from the primate mind during the earlier stages of human evolution. As for the rest of species, he considers that they can show some relatable characteristics as reflexes, instinct, curiosity, behavior, and memory, among others. However, his thesis does not support the idea that there is any continuity of these aspects from the less complex organisms to the most complex ones.

From a genetic approach, , an evolutionary biologist and geneticist who opposes genetic determinism, explains how although genetic mechanisms are usually

8 considered in the study of behavior, the precise paths of such genetic mechanisms for cognition are still unsolved. One assumption of his work is that, instead of the analysis of the biochemistry of genetic mechanisms in cognition, we could study these changes through Natural Selection. He also argues that the evolutionary questions of cognition are not related just to the evolution of cognition itself, but also to the effects of cognition on evolution (Lewontin, 1998). Thus, he became a pioneer in this track of the study of cognition.

Moving forward, we mention the philosopher and linguist Karen Neander, who introduced a different definition of cognitive systems: “systems adapted for producing and processing internal states that carry information, and for using these states to adapt the bodies in which they are situated to the environments in which they, in turn, are situated and vice versa”. From her perspective, there is indeed a distinction between mind and brain, and moreover, the environment will also play an independent role (Neander 2007).

For Neander and Dennett, cognition is a characteristic that can be found in all organisms. For example, they support the notion that organisms such as plants or fungus can in fact receive information from the environment, process it, and respond to it. This issue had previously been questioned by philosopher of biology Peter Godfrey-Smith, who coined the term “proto- cognition” in order to refer to the cognition of plants and bacteria; “capacities for controlling individual growth, development, , and behavior by means of adaptive response to environmental information” (Godfery-Smith, 2002). However, two objections can be made: firstly, this description of proto-cognition does not take into consideration evolutionary changes, and secondly, the definition implies that the proto-cognitive characters may well be seen just as an extended part of behavior or development.

Based on the controversy that the differentiation between “proto-cognition” and cognition was causing, the American philosopher Hilary Kornblith tried to come up with a solution in 2007, arguing that the term “cognitive organism” should be used for those organisms that not only receive and process information, but that can also make a representation of such information.

Dieguez tried to find a different solution to this problem and thus divided the organisms into two different categories: on one hand, organisms with internal representations, and on the other, organisms with mental representations. The first group of organisms is comprised by those that, after a stimulus, are capable of responding, and such a response may or may not cause a deterministic change of behavior. Internal representation is something that all known types of organisms present during their life. For the second kind of organisms –the

9 organisms with mental representations– the representations have a neural basis, and can be evoked without a stimulus: both the production of a new memory and the remembrance of an old one can activate neural activity. Thus, this kind of representation is found only in animals with a complex nervous system (Diéguez, 2011).

Horik and Emery, a duo of psychologists, analyzed the possibility of a link between cognition and specific aspects of species’ lives. In this case, cognition is not related only to the organism itself, but to sociality, culture, tool use, and behavioral flexibility, among others. As such, cognition influences more than just one environmental selection pressure at a given time. It is important to mention that Horik and Emery side with the view that all organisms possess cognitive characters. For them, it is likely that different species who shared analogous environmental selection pressures evolved with similar cognitive abilities. This explained the wide variety of cognitive characteristics and, at the same time, implied that all animals share many fundamental cognitive abilities, but with a different development of them (Horik & Emery, 2011).

Recently, Dennett offered a description of the origin and evolution of cognition with a phylogenetic and functionalist angle. From his perspective, NS is the main protagonist of the story. His explanation goes from the origin of cognition in the first replicating organisms, to the origin and evolution of the nervous system and its function. He stresses the importance of brain plasticity for cognitive abilities such as learning. In this way, Dennett's arguments lend themselves adequately to the treatment of how a phylogenetic and adaptive account of cognition is constructed. Due to this, in the following section we focus on Dennett’s theory.

1.2 Cognition: an adaptationist property according to Daniel Dennett

In 1993, Daniel Dennett described an adaptationist approach of cognition in his work titled “”. His work, opened a new tendency in the comprehension of cognition under the Ultra-Darwinism scope, explaining that cognitive properties, as consciousness, are also subjected to natural selection processes. In order to support his work, he proposes a phylogenetic hypothesis of the phenomenon by which cognition could give rise, describing a series of events that could occur through populations of different organisms. In the next section we will explain in detail his hypothesis.

10 1.3.1 Evolution of simple entities by Dennett

For Dennett, the extremely rudimentary lifeforms are called “replicators”, and probably the earliest ones in the history of life on this planet, were even simpler than actual viruses (Zawisizki, 2014). Then, for the simplest replicators, the only way to continue replicating, the replicators had to take the good things, repel the bad ones, and ignore the neutral ones that the environment offered. (Dennet, 1993).

The variability of the replicators was given by faults during replication. The copies of faults propagated, and the ones that survived were the ones that presented a higher fitness2. In these terms, we will be talking about NS and , given by a genotypic variation. Thus, the evolution by NS will lead to particular types of changes in the population of replicators, given by the maintenance of the characteristics that increase fitness.

According to Soberón, an ecologist interested in evolutive biology, in the study of evolution there are a series of properties that an entity has to have in order to be considered as an “Evolution Unit” (EU): i) the replicator has to have a phenotypic variability, ii) at least a part of that phenotypic variability has to be heritable, iii) the heritable variability could be related with the probabilities of the survival and replication of the replicator. Soberon claims that if an entity meets the first two requirements, then it can be considered as an EU, and if all three conditions are met, then the entity can be considered as a Darwinian Evolution Unity (DEU). Given this, according to Dennett's description of replicators, the simpler replicators were DEU and EU.

1.3.2 From simple replicators to the nervous system

The nervous system is a complex and specialized system. In order to generate such system, it is necessary the association of specialized cells. These cells, have evolved to respond and discern to external stimulus. In 1993, Dennett suggested that “protoneurons” evolve through NS, developing an unstable membrane potential, that propagated into other population of cells. After, the diversification of was given by morphological changes as the increase in the size of dendrites and axons, giving rise to a more complex network. In that sense, this also gave place to the specialization of neurons, in which, some of them will focus on the processing of information, or work as motoneurons, among other things (Angrino, 2010).

2 “In the crudest terms, fitness involves the ability of organisms— or, more rarely, populations or species— to survive and reproduce in the environment in which they find themselve” (Orr, 2009)

11

The evolution of simple entities into networks gave place to three different types of nervous systems that we find in the present phyla. According to van-Wielink, a neurologist specialized in neurodegenerative diseases, the first type of neural system is constituted by an nonsynaptic net of cells that is present in less complex animals, as cnidarians and echinoderms, characterized by single units that communicate each other by calcium waves and other impulses, to control simple actions. In contrast with this, there are some authors that postulate that this can not be considered as a nervous system, instead, this will represent the most primitive system, being considered as the precursor of the nervous system due to its lack of synaptic junctions and less specialized cells in contrast with neurons (Jacobs et al., 2007).

The second category is characterized by the presence of ganglion. This system is found in animals that represent the next stage in evolution, in terms of complexity, as arthropods and annelids. In this system, the neurons are no longer single units, in contrast, they build a segmental ganglion, that as the name says, will modulate a segment of the body, and will be connected to the next ganglion by “connectives”.

Finally, the last category is the most complex and its found in . It is characterized by the presence of a neural tube and the cephalization of the system. It is divided into central nervous system and the peripheral nervous system. The communication between cells can be by electrical synapses and chemical synapses, and/or extrasynaptic release. This nervous system gives rise to more complex systems in terms of intelligence.

Dennett (1993) makes evident that for primitive animals with a simple nervous system, the signals from the environment were innately modulated. As for what he called “proximal anticipation” behavior, to respond into an immediate future, and “short-range anticipation” behavior, to the capacity of an animal to produce a more elaborate response than a reflex. He states that these are really flexible, and so, be consider as highly adaptive characters, that contributes to the fitness of an organism that can be extended into populations.

Despite Dennett's functionalist approach, to explain the primordial behavior of the first animals with a nervous system, he only tells us that they are equipped to solve ecological problems. The cognitive tools that appeared since very early and that we all possess now, like the reflexes, the orientation and the capacity to recognize objects, are only one face of the currency of what can really be deduced from the origin of evolution of cognition.

12 Without being able to explain the ecological contexts of selective pressures, whether a niche is chosen or created, whether they were only instinctive characteristics or manifested as emotions, among other things.

1.3.3 Phenotypic variation and the understanding of cognition

Following Darwin's theory, the design that animals own is not always the most fitted for a specific environment. However, the ones that are able to redesign will be the ones that will survive and reproduce. This means that there is a certain level of variation to which we can resort to our life depending on the eventualities that arise. This characteristic is called “phenotypic plasticity” (Dennett, 1993).

Thus, in terms of cognition, natural selection has been responsible for designing cognition, acting on the variations of nervous systems that have existed, leaving only those who responded effectively to environmental interactions and who could inherit the characteristics that helped to survive their carriers.

Dennett claims that when the “plastic brain” is exposed into novel things in its environment, the brain reorganize. The process by which this happens is by a similar process than natural selection. Everything starts in an individual brain by postnatal fixation. In this case, the brain structures that are select, will be the ones that can control or influence behavior. The mechanic process of elimination will lead selection, that at the same time, it has a genetic background. In this case, the organisms with brain plasticity will have an advantage over the ones that does not, and this might accelerate evolution by NS.

For Dennett, the “good tricks” that an animal learns during its whole life are going to improve the animal fitness. This claim, gives a lot of things for granted, and specifically, not all the good learned tricks will lead to a higher probability of reproduction, and therefore the fixation of the character, will not occur. For example, a can learn how to use a tool to take the termites from their holes. However, this does not assure him that he will become the alpha male of the herd and then have more chances to reproduce. In response to this issue, Diéguez (2011) proposes that natural selection acts on individuals with a tendency to have certain cognitive abilities. The selection would have been produced by the disposition to learn them. Thus, learning the “good tricks” would have easily passed the next generation in a non-genetic way, however, what is genetically passed is the disposition to learn them.

13 In summary, Dennett's evolutionary explanation tells us that cognition is a product of natural selection, acting on the variations in the different nervous systems that are responsible for producing cognitive phenomena. Animals that have cerebral plasticity, are those that have been able to evolve to a "more complete" cognition. The complexity of cognition leads to those that apart from the recording and processing of information, also have a representational system of the information that comes from the environment, that is, they may be able to have beliefs and desires about the environment. Finally, with cerebral plasticity, new neural connections can be formed, triggered by the experiences that individuals have throughout their lives, thus contributing to their survival and reproduction.

As we could see, cognition is diverse, but is not impossible to find a common type of evolutionary story that could apply to most of the cases. Nowadays, cognitive and behavioral aim to study the brain and its functions using a wide range of methods. Within the most popular techniques to record the brain activity we can find: electroencephalography, magnetic resonance, electrophysiology, among others. These techniques are use in order to understand mechanisms such as motor behavior, thought, consciousness, memory, perception, language, among others. At the same time, other sciences as the philosophy of mind, the ecology of behavior, cognitive paleoanthropology, , and are using their own scientific methods to complement the origin of the evolution processes of cognition. To postulate a multilevel causality for the understanding on the adaptationist idea of cognition, first we need to introduce the processes that we would like to use. So, in the next chapter I will present the first process, ontogeny.

14 2 Ontogeny and the “theory of neuronal group selection” in the understanding of cognition.

2.1 Introduction

Ontogeny is the origination and historicity of an organism, from egg fertilization till its death. Along with its life, the organism will undergo physical and behavioral changes that can be due to external factors (environmental interaction), or internal factors (). In 1977, Stephen J. Gould, one of the most influential evolutional biologist from the XX century, stated that evolution will occur when ontogeny is transformed. This transformation can be due to an introduction of new characters or by the change of an old character. Following this, the change of an old character will have a regulatory effect, that will change the rate for features already present, and the introduction of a new character might give rise to a new feature.

The understanding of how cognitive capacities are build giving the present brain structures, it is essential in the study of cognition. In 1987, Gerald Edelman, a biologist winner of a Nobel Prize due to his work on the immune system, proposed a theory to apply Darwin’s theory to the evolution of cognition, in which, he claims that all cognitive functions have a biological support that lay in the brain. Indeed, those properties will be regulated by NS and will apply to all organizational levels from biochemistry to . In this theory, he applies NS on a different level than Dennett. In this case, Edelman speaks about a population of cells instead of populations of individuals. In this chapter, I will like to explain and try to relate Edelman’s evolutionary explanations about the cognitive properties that might complement the understanding of cognition under an adaptationist scope.

2.2. Edelman and the variation and selection within neural populations

The theory proposed by Edelman called “” or “The theory of neuronal group selection” (TNGS), tries to fill the gap between biological bases and . All started with the necessity of merging two different observations of brain function.

One of the observations was the structural and functional variability of an individual nervous system. According to Edelman (1987), the variability will be present at a molecular, cellular, anatomical, physiological and behavioral level in time and space. To explain this variation

15 between individuals, even from the same species, Edelman claims that the adaptive characteristic will emerge in the course of an individual development (ontogeny).

The second observation relies on the understanding of the development of such adaptive characteristic and its relation with the world stimuli, i.e., “in order to survive in its econiche3, an organism must either inherit or create criteria that enable it to partition the world into perceptual categories according to its adaptive needs” (Edelman, 1993).

Furthermore, Edelman argues that the ability to categorize a novel input from the environment and respond to it, from an adaptive perspective, comes from processes of selection upon variation, instead of the presences of a semi fixed neuroanatomy that just reads the manual of instructions in order to respond and adapt (Edelman, 1993).

In this case, if no individual result of sexual reproduction is identical, then, no two are alike. Each brain has its own developmental process and is constantly changing during its life spam. Edelman and Tononi (2000) point out that during the process of natural selection the phenomenon of correlative variation can occur where a “primary trait can be selected for and bring along another changes that are used later for other selective events” (Edelman and Tononi, 2013). For example, selection of an enlarged brain structure to facilitate perception may be accompanied by the enlargement of other neighboring regions in the brain and subsequently these regions may be selected to perform another function, such as memory.

2.2.1 The three main tenets of the theory of neuronal group selection. The Neural Darwinism theory is built under tree main tenets explained below:

1. Developmental selection. Darwinian natural selection and evolution is usually studied in populations of organisms, but when it applies to cellular populations; is called “somatic” evolution. Such somatic evolution tends to reduce cooperation among cells, thus threatening the integrity of the organism (Edelman 1994). Given the above argument, genes and inheritance give the formation of an initial of the brain. However, the connectivity of synapses is established by the somatic selection during each organism development.

3 The econiche definition stills controversial, however, in this work we refer to econiche to “what describes a species’ ecology, which may mean its habitat, its role in the ecosystem, etc” (Pocheville, 2015).

16 As an example, during neurogenesis, dendrites and neurites will mature, giving rise to new branches that grow in several directions. This will give rise to new patterns of connection, which in turn will produce a vast and varied repertoire of neural circuits. Then, according to neural individual patterns of electrical activity, neurons will create populations, ending in a system in which “neurons that fire together, wire together”. As a result, neurons of a population will be more closely associated to each other that to neurons in other population. (Edelman and Tononi, 2013).

2. Experiential selection. The process of synaptic selection will occur inside the repertories of already existing groups of neurons due to behavioral experience. For example, the maps of the brain that corresponds to a finger response can change their confines, depending on the use of them. When a finger is trained to be used in a certain way, then, the synapses between populations on charge of this response will get strength, without any anatomical changes (Edelman and Tononi, 2013).

3. Reentry. This tenet allows the integration of the previous two tenets, leading to the synchronization of the activity of groups of neurons in different brain maps, transforming them, temporarily, into a big circuit with a coherent output. The ability that we have to discern between movement and shape in a display of moving dots, due to the integration of different brain areas, is an example of this (Edelman and Tononi, 2013). The reentry of connections between neuronal groups in diverse parts of the brain by a single stimuli sense by different senses, will coordinate the impressions from all the senses to provide a coherent, consistent, continuous experience. The reentry of information also will provide a mechanism of re-categorization4, the fundamental process by which the brain carves up the world into different things and recognizes those it has encountered before.

4 The word re-categorization, it is not to be taken as implying the existence of a prior set of categories: in fact, every act of recognition modifies the category.

17 In this theory, Edelman stress out the importance to high-order processes -as thinking, planning, perceiving, and language-, in which concepts are maps of maps of the brain, which will get rise from the re-categorization of the brain activity. The first –order of consciousness will be given by the concepts by themselves, and in human consciousness, we will also find the features of a as the concepts about concepts, language, and a concept of the , that will be built on the foundation of first-order concepts.

In summary, in terms of cognition, some patterns or population connectivity’s will be reinforced by experiences, while many others will be eliminated in a selective process. Some other type of research supports this theory, as the consolidation and reconsolidation of memory (Edelman y Tononi 2000; Nader et al., 2000). Neural Darwinism attempts to explain how some cognitive abilities emerge at the cellular level. Neural Darwinism is an incomplete analogy to the historical process of Darwinian natural selection. The difference is the time scale and the selection units. In this theory, the entities under natural selection are the neural groups. The first stage of selection occurs during the of the individual. This first pattern of connections between neural groups are those given by value systems. The second stage is the selection given by experience, which occurs during the postnatal stage and until the death of the individual. Contact with the environment creates modifications between the different connections of the neural groups with degenerative characteristics and as a consequence gives a flexible or plastic property to the neural groups. Finally, the process of re-entry is all the reciprocal connections that are distributed throughout the brain, which marks the coherence to produce the cognitive phenomena.

This whole repertoire known as neural Darwinism is the basis for understanding how cognition develops in each individual. There are other elements that complement this theory and which I consider below.

2.2.2 Degeneracy and value

The theory argues the existence of another essential and unique property that all the selective systems will share, denominated “degeneracy”. This property refers to the ability of structurally different variations of brain elements to produce the same function, i.e., “many different ways, not necessarily structurally identical, by which a particular output occurs” (Edelman and Tononi, 2013). For Edelman and Tononi, this property will occur at one organizational level or across a multiple, otherwise, all would be lethal.

Finally, another key idea in the theory is value, a word used here to describe inbuilt tendencies towards particular behavior. These forms of behavior may be driven by what we

18 value in a fairly straightforward sense - seeking food, for example, but they also include such inherent actions as the hand's natural tendency to grasp.

2.3 Edelman’s theory applied to the immune system

Edelman constructs an analogy of TNGS with what happens to the immune system. In this case, instead of population of neurons, the entities under NS are the . He proposes that from a vast variety of antibodies, the ones that will be selected to rapidly reproduce will be the ones that linked successfully to the target molecule. That at the same time, increase the fitness of the organism. In this case, the immune theory, from which Edelman formulates his analogy, postulates that there is no possible way that the body will have all the antibodies for all the foreign substances that could attack an organism. In contrast, the body will rapidly produce an for substances it has never encountered before (and indeed for substances which never existed in the previous history of the planet). In an analogous way the TNGS explains how the brain can recognize objects in the world without having a huge inherited catalogue of patterns.

On the contrary, the biologist Steven Rose (2001), does not agree with the name granted by Edelman of neural Darwinism. Rose believes that it is neither a homologous process nor sufficiently analogous to Darwinian natural selection. Considering neural Darwinism only as a metaphor (Rose 2001) Even Rose agrees with in calling it 'neural edelmism'.

Rose's strongest argument against neural Darwinism is that overproduction and subsequent selection of neurons and synapses, is actually a cooperative process rather than a process of competition. Migration of cells and axon growth over long distances is only possible by remote and local signals between the same neurons. On the contrary, if those signals are not present during the growing and migration period, then, the axon wouldn’t have reached its final destination. As far as cells do not "compete" between them, it can rather be seen as a "cooperative".

19 3.Evo-Devo: an extension to cognition

3.1. Introduction

As we saw in the last chapter, when it relates to cognition, ontogeny does not actually recapitulate phylogeny, as Haeckel proposed in his “recapitulation theory” 5 (Gould, 1997). Now, it is necessary to discuss the relationship between embryological development and evolution of cognition, since embryos also evolve in different ways, putting the recapitulation theory seen as a historical side-note, rather than as a dogma in the field of developmental biology. Then, to understand the relationship between embryological development and evolution of cognition, we will explore the interdisciplinary field of research called Evolutionary developmental biology, also known as Evo-Devo.

Evo-Devo studies the differences between organisms during their development to determinate their phylogenetic relationships. Evo-Devo offers explanations about the phylogeny based on the heritability of their altered characters produced during the ontogeny of an organism. At the same time, Evo-Devo claims that the design of a character will not be totally predetermined by the . In these terms, genes are not the only causal agents of development. At the same time, each developmental stage will react to multiple interactions between environmental and epigenetic factors. Evo-Devo does not question NS as an evolutionary force, but also delimit to what degree the NS can be considered as a creative force. When it relates to NS, one of the most controversial topics is the position of NS as a creative force or not. Against this position, authors such as Gould and Lewontin, argue that NS does not “create” features, adaptations, or even life. For them, NS it merely selects a feature that provides greater survival rates. On favor of this position we can find authors such as Ernst Mayr, who claimed NS as "an all-powerful natural selection".

Furthermore, all the processes related to the development of the brain plays an important role in the evolution of cognition. An Evo-Devo approach will allow us to specify how much weight developmental constraints have on evolutionary models of cognition. Recognizing that these constraints do not just limit, but also order the evolutionary path producing a bias of

5 The theory of recapitulation is commonly synthetize in a phrase given by Ernst Haeckel' as "ontogeny recapitulates phylogeny". This refers to a questioned biological hypothesis that argues that during the embryological development of an animal, from fertilization to gestation, i.e. ontogeny, all the successive stages from which the animal undergo represents stages in evolution of the animal's remote ancestors (phylogeny).

20 evolutionary phenomena that which cannot be explained by explicitly citing the classical evolutionary factors (eg, NS).

3.2 Evolutionary developmental biology theory

Nowadays, evolutionary developmental biology (Evo-Devo) theory studies: i) the evolution of development, by a comparative approach of features at different hierarchical levels (Hall, 2003); ii) the establishment of homologies, by genetic expression patterns (Amundson, 2005); iii) evolutionary innovations, to comprehend the mechanisms that give rise new characters (Hoekstra & Coyne, 2007); iv) phenotypic evolutionary patterns, to determinate if development constraints can influence evolutive diversification (Müller, 2007); and v) genotype- mapping, for the understanding of the dynamics of adaptation (Pigliucci, 2010).

However, the study of Evo-Devo as a discipline started in the 1980’s century, with the discovery of the homebox. In 1983, Walter Gehring from the University of Basel and Matthew Scott and Amy Weiner form the University of Indiana found a highly conserved 180-base pair (bp) DNA segment in animals, fungi and plants, involved in morphogenesis and which encodes a polypeptide segment designated as homeodomain (Gehring 1992). Later on, during the study of the development genetics of the fruit fly Drosophila, researchers found a coding sequence involved in segment formation. This sequence was found later along different species in clusters of related genes, now known as hox genes.

The study of homologous genes in different species ended up by sealing the connection between genetics, developmental biology and evolution. Despite the great conservation of genes, the morphological diversity between taxa is given by the variations in the expression of these genes. These variations would produce along the development morphological and functional differences between taxa on which NS would operate adapting the organisms to the environment and generating the enormous biodiversity past and present (Baguñá 2003). With this, scientist could not avoid the study of developmental genetics in terms of evolution, giving place to “Evo-Devo” (Carroll, 2008).

Evo-Devo theory claims that the genotype does not completely define the design of an organism, and it considers that mutations have a strong effect on fitness (Carroll, 2008). Within this theory, genes are not considered as the only causal agents of development. Moreover, the role of genes in development and evolution does not lie above that of the other

21 factors involve in such processes. Evo-Devo supporters believe that environmental and epigenetic factors bear equal importance as genes in the development of an organism (Balari & Lorenzo, 2008).

3.2.1 The neural system from an Evo-Devo perspective.

As we mentioned above, Evo-Devo claims that there are patterns involved during the development of organisms. However, these patterns are also found in the development of the nervous system. Nowadays, research has revealed the existence of different stages of brain formation, starting from a simple set of undifferentiated cells that grow and become the neural plate, that will fold to form a groove then tube, open initially at each end. Three projections come out of this tube. The first is the one that will give rise to the series of segments that make up the rhombencephalon, which will contribute to breathing, balance and keeping the body alert. The second segment will give rise to the mesencephalon, which coordinates additive and visual reflexes. Finally, the third will give rise to the precursor of the forebrain, responsible for reason and decision making (Nicholls et al., 2012).

In order to develop a nervous system, brain cells must specialize and migrate to their final locations, otherwise this will cause irreversible damages to the organism. Initially, only certain regions of the body are responsible for manufacturing cells in an "emergency" case, where by means of a cellular Darwinian strategy, those that have been able to integrate into a previously established and large system will be selected (Marcus, 2005).

Gene expression plays a very important role in terms of complexity, since they establish and adjust the neural circuits. So, we could say that the brain structure, as a cognitive substrate, is the result of transformations that occur during the development of an organism. Where the constrains6 of brain development enter the field of study of Evo-Devo. With this approach, it will be possible to determine the weight of the causes of developmental constraints in evolutionary models of cognition such as the adaptation, described by Dennett, and ontogeny, described by Edelman. Due to the importance in the understanding and definition of developmental constraints in evolutionary developmental biology, we will revise in detail the constraints in the next subsection.

6 “Another consequence of interacting modules (in terms of the homebox) is that these interactions limit the possible that can be created, and they also allow change to occur in certain directions more easily than in others. Collectively, these restraints on phenotype production are called developmental constraints”.

22 3.2.1.1 Types of constraints

Evo-Devo recognizes that the constraints and directions not only limit but also order the evolutionary path producing a bias in evolutionary phenomena that can not be explained by citing exclusively the classical evolutionary factors (Caponi 2009). It is inevitable to think that they are relevant to the evolutionary process. They are relevant in the sense that the alterations in ontogeny are produced during the same ontogenetic process and changes in this process can generate a totally viable product (Amundson 2001). In this way, these alterations are the ones that produce the variations in which later the natural selection will act. I will rely mostly in the work of Gilbert to classify them, as: the physical constraints and phyletic constraints.

The physical constraint relates to laws of physics as the laws of diffusion, hydraulics and physical support, and chemistry. Combining these laws, the physical constrains build rules of how these laws influence directly the processes of development of all organisms, without any exception. For example, there is no way to have a on wheeled appendages, because the blood will never irrigate these organs due to the fluid dynamics law that establishes that blood could not flow in a rotating organ. Another example is the elasticity and tensile strengths of a tissue, that is use in division, migration, apoptosis, and shape of a cell, that will limit the size of the cells (Gilbert, 2000).

The phyletic constrains are the ones that give identity to a taxon, as they arise as a consequence of some particular characteristic fixed phylogenetically in a group. Because of this, the developmental path will be more limited, i.e. once a structure come to be generated it is difficult to reset it and start over again from scratch (Gilbert, 2000; Martinez 2013).

This term of “constraint” in literature has never been used as a neutral evolutionary force, but has been attached both a negative and positive weight, depending on the author and the case. When we talk about a constraint with a negative weight, it refers to the fact that it will limit or restrict the variation on the number of possible phenotypes in which the NS can act, thus being a quantitative aspect. On the other hand, when a negative weight is passed, the aspect will be qualitative, in which the role of constraint will print the evolutionary direction to an organic form since it will only drive the development of an organism by specific path (Martinez, 2013; Caponi, 2009).

In the following section I will explain how all the factors of Evo-Devo mentioned during the chapter, can modulate cognition.

23 3.3 Evo-devo and cognition

In terms of constraints, Parker et al., proposed that constraints can have a significant effect on the cognitive capacities of different taxa. When comparing the brain development of Rhesus monkeys and humans, they found that the myelination of the extends in humans up to 12 years of age, whereas in the Rhesus occurs until the 3.5 years. In turn, dendritic growth in humans occurs up to 20 years, while in any other species, this growth ceases to occur long before. These factors could influence the cognitive differences between monkeys and humans, proving that the variations occurring during development are those that limit the brain phenotype of each individual, and therefore, the range of variation.

In contrast, Emery and Clayton (2004) observed that some birds and primates have cognitive similarities and do not have the same brain. However, the similarities might rely in the idea that both species have faced similar adaptive problems, such as locating food in a wide temporal or spatial distribution. In turn, both species use simple tools, have a very developed imagination, and flexible behavior. So, how come your brains differ so much? The authors claimed that this could be due to a possible establishment of homologies based on the expression of their regulatory genes on brain development.

As we saw earlier, Dennett claims that all variations in nervous systems arise from mutation. On the contrary, modern synthesis tells us that genetic recombination and mutation are the evolutionary forces. However, the laws of the Evo-devo, includes regulatory genes as a very important factor in evolution. Regulatory genes are responsible for the activation or inhibition of expression of other genes. For example, genes such as Otx and Emx, that control the balance between the mesencephalon and roboencephalon, and which determines the balance between the hippocampus and the frontal cortex, respectively (Marcus 2003).

In relation to regulatory genes, Marcus offers examples of these regulatory genes, noting that the evolution of neural specialization did due in part to random genetic duplications. However, Lewontin, criticizes this position, explaining this explanatory scheme as a bad biology. According to Lewontin, the genetic variations that depend on mutation processes are very rare phenomena, since from millions of probabilities only one would have the possibility of fixation. Moreover, there is a very low probability that this mutation appeared in the next generation. All this without considering the time it takes to fix a character in a population under NS. For Lewontin, the phenomena of cell migration or experience are the processes that generate variation in the development of the nervous system using Edelmans’ theory.

24 On the other hand, Balari and Lorenzo (2008) argued that, anatomical structures in specific brain areas of different individuals of the same species are very similar and their development must obey genetic constraints of some kind. There is a high degree of diversity in terms of neuronal morphology and neural patterns. Such variation can only be derived from epigenetic factors during the development process. As new factors of variation in the synaptic level are introduced, they will change the biochemical structure and the origin of an increasing number of different neurotransmitters. Therefore, a process of this type has an enormous potential for introducing novel morphologies, with obvious repercussions at functional level.

In conclusion, the arguments in favor of alterations of the cerebral development with importance in the evolution of the cognitive capacities is getting attention. Variation and diversification of species enter into reconsideration because new causal agents of development have been found. The mutation and NS can no longer be seen as the only responsible fountain of variation. But the connection between and phenotypes is complicated and diverse.

25 4. Multilevel causality models and cognition

4.1 Introduction

In the previous chapters we explained Dennett’s adaptationist theory, and the ontogenetic and Evo-devo theories of cognition. The explanation of the theories provided the basis to build an integrative adaptive theory of cognition, helping us to develop a model of multilevel causality based on ontogeny, evo-Devo, and NS. In this chapter we will summarize the evolutionary theory given by Darwin, from where adaptation gives rise. Also we will focus on the discussion points of why each process can’t give an adaptationist approach of cognition by it self, followed by the types of multilevel causality models, to finally build a model to merge these theories for a better understanding on the adaptationist theory of cognition.

4.2 The understanding of adaptation

Natural selection is the only evolutionary processes that produces adaptation. The term of adaptation is disrupted and still controversial due to the differences in beliefs among current experts in the field of evolution. Different authors have concluded that there are at least three approaches to adaptationism: the empirical, the explanatory and the methodological approach (Orzack and Sober 1994, Sober 1996, Amundson 1988, 1990, Godfrey-Smith 2001). However, in the present work we refer to “adaptation” as the development of any behavioral or genetic character that will allow the organism to cope with environmental changes to survive and reproduce (Smit et al., 2006).

Natural selection acts on phenotypes, but evolution (as it is classically understood) consists of changes in allele frequencies. Since evolution is the response to selection, this will only occur when the selected characters are inherited and therefore, have a genetic basis.

Nowadays, there are explanations that emphasize that heredity surpasses the genetic level. For example, Jablonka and Lamb (2005) proposed four different types of heritage with evolutionary relevance as: i) genetic inheritance is the same mechanism that postulates the modern synthesis; ii) epigenetic inheritance is the cellular self-organization that has effects in the phenotype such as: DNA methylation, self-sustaining loops, RNA interference, and chromatin systems labeling; iii) behavioral inheritance such as: transfer of substances influence behavior and imitative and non-imitative social learning and, finally, iv) symbolic

26 inheritance is the language that only appears in human beings, and is partially fixed in an innate way but is learned socially.

4.3 Difficulties in the explanation of cognition by the adaptationist approach.

4.3.1 Difficulties in Dennett’s approach and ontogeny

During the second chapter, we examined Dennett’s arguments. He claims that in all types of animals, including humans, cognitive capacities are a phenotypic trait. This phenotypic trait will result as an environmental adaptation, i.e. NS. In summary, what Dennett tries to say is that cognitive abilities are passed on to the next generation in a non-genetic way. On the contrary, what is genetically transferred is the disposition for learning. Thus, non-genetic inheritance, such as experience or culture, is made possible by genetic inheritance, i.e. the plasticity of the brain, with readiness for the experience.

On the other hand, what modern synthesis does not say is that NS will act on a variant of individuals, however, the impact will be on the population. When individuals from some populations were selected because of disturbances in their region or differential reproduction, it was observed that the selected individuals did not change at all, but simply survived the selection effect or were further reproduced, while others died or reproduced less. What changed after the selective process, were the characteristics of populations, not individuals (Freeman & Herron, 2002). For example, lets suppose that we have a population with four variants (A, X, Y and Z), and that each variant has a 25% proportion in the population. Then, after a selection process, the proportion of the population changes. If we assume that only the variants A and X survived, then the new proportions of the population will be 50% and 50% of each variant. In this case, variants Y and Z did not change either their genotype or their phenotype, instead, there was a change in the total population variation.

Applying the previous argument, related to cognitive traits, the adaptational explanation given by Dennett, would have to be reconsidered. That is, when an individual is subject of natural selection, the phenotype doesn’t change. Individuals who face riots in their niche, whether triggered by an external agent or by themselves, also result in their cognitive structure. The plastic capacity of the brain makes possible that this can be restructured throughout the life of the individual. Then, we could assume that changes occur in their phenotype, and in this case the change is not heritable because it has not been in the genotype. However, brain plasticity does. Each generation is the product of selection by the conditions of the niche that prevailed in the previous generation. The ability of natural selection is clearly seen looking

27 backwards, not forward (Freeman and Herron 2002). Descendants of individuals with cognitive abilities that aided their survival and reproduction are better adapted to the niche in which the paternal generation lived. However, if these conditions change again during the life of the offspring, they would not necessarily be adapted to the new conditions. Other considerations need to be taken into account when populations also create their own niches. Lewontin (2000) points out that organisms determine what elements of the external world are going to constitute their environment and what relationships between those elements are relevant to organisms. In addition, they not only determine what aspects of the outside world are relevant to the characteristics of their form and metabolism, but actively build and modify the world around them. Thus we can say that individuals also inherit their niche.

Although natural selection is continually optimizing adaptation, evolution does not give rise to perfect characters. In Chapter 2, we looked at Van-Wielink's (2002) proposal on the three different types of nervous systems and the types of intelligence that correspond to each. The correlation between the form of organization of the nervous system and the character - in this case understood as the intelligence - is due to the fact that the same disposition of the nervous system is the one that sets the stage for the development of a type of intelligence and no other. Therefore, NS gives rise to different types of nervous systems with types of intelligence that may present, even though, in rigorous terms, NS should just act on favor of animals with higher intelligence skills. This is because the selection does not optimize all the characters involved, it only leads to adaptation, but not to perfection. So the selection is neither progressive nor random. It is not random, since it increases the adaptation to the environment. The non-random process is completely free of any conscious intentionality. Actually Darwin even rejected the coined phrase "naturally selected" because people believed that the word selection involved a conscious act or choice by some being (Freeman & Herron, 2002). However, none of this happens. It can only be said that there is an increase in complexity, degree of organization or specialization. It can not be said to be progressive because there is no inevitable tendency towards more “advanced” forms.

A further aspect of natural selection is that it can produce new characters, even when it acts on existing characters. That is, NS could only act on the basis of variation that already exists in the population (Jacob, 1977; Freeman and Herron 2002). Also known as creative natural selection, and with time, the selection could give rise to new characters. For the understanding of the creative action of selection, Martínez and Moya (2011) merged two concepts previously described by two authors: canalization and bricolage. The first one proposed by Dobzhansky (1973), mentions that the variation is directed by its historical load. The variation by mutation and recombination is not random since it is conditioned by its

28 evolutionary history, rather, canalized, by the functionality it had before the mutation and recombination, and by the same action by the natural selection for a new character. The second proposal is by Jacob (1977), and refers to the bricolage of selection. Broadly, it tells us that selection is capable of "reorienting" behaviors, structures or genes from existing material.

To sum up, the of NS leads us to support an explanation with a strong adaptation component for the mind, stating that some cognitive capacities emerge as an evolutionary adaptation to a variable and complex physical environment. Such capacities, as the perception, the representation of the environment, memory, and learning, among others; play an important role in survival and differential reproduction of organisms. Cognitive abilities arise to deal with complex environments, whether ecological or social, through flexible behaviors (Godfrey-Smith 2001). However, we must keep in mind that the adaptive hypothesis of the origin of cognition is not the only explanation that exists. Taking into account the article written by Gould and Lewontin (1979), in which they highlighted that all explanations about evolution by NS always lead to an adaptational argument, and that adaptation is not the only and most important explanation to which it can be appealed evolution, adaptations under the scheme of modern synthesis, are passive adaptations always at the expense of the environment. This is something that similarly happens with the canonical idea of the origin of variations only by mutation and genetic recombination.

However, throughout chapter IV, we developed other arguments that may serve not to refute the modern synthesis, but, to my way of seeing, to complement that theory. In this case, mutation, recombination and natural selection are not the only causes of variation. Instead, there are also constraints on development that have evolutionary repercussions on cognitive processes. On the other hand, natural selection does not act on passive individuals, the cognitive capacities that the organisms carry makes that the interaction with the environment is not merely restrictive, but what is observed is a feedback. Then, if we use an adaptive explanation of cognition, that takes into account more causal factors such as development constraints and imposes limits on natural selection. Then, we will be able to build a more accurate model of how cognition evolves in animals. This idea has been characterized by the biologist Franz Wuketits (2006) as an anti-adaptational approach. This approach denies that there are passive adaptations to the environment, that is, because organisms modify their environment, and this, once modified acts on the organism. The main idea is that cognition is the function of active biosystems, and not blind that only respond to the outside world. Therefore, cognition is not a reaction to the outside world, but results from complex relationships between the organism and its environment.

29

Finally, cognition is not a linear process of accumulation of information step by step, but a complex process of continuous elimination of errors (Diéguez, 2011). The anti-adaptational approach is not really an alternative to the adaptationist, since it is limited to emphasizing the active role of the organism in the transformation of its environment. In contrast, it highlights that the medium has also been transformed by these cognitive capacities and that in the evolutionary process, the constrictions of various types and the internal self-organizing factors have played an important role (Diéguez, 2011).

4.3.2 Analysis of Edelman and Dennett’s theories: differences and similarities.

Apparently Dennett and Edelman share some crucial views about neuroscience. They both consider a "mechanistic" model as heuristically fundamental for the sake of a scientific theory of consciousness, and support an ultimately naturalistic ontology. However, there are crucial differences between their theories, in which Dennett’s theory is classified as a “multiple drafts” theory and Endelman’s theory is a “dynamic core” theory.

According to Dennett, the distributed neural/ cognitive models will produce content in parallel and the system will receive a really important impact from the conscious content (Dennett, 1991). To explain this, he use the metaphor of “fame in the brain”, in which there is not precise time at which there is no precise time at which particular content becomes “famous” and fame can only be determined retrospectively (Dennett, 2001).

In contrast, the dynamic core theory argues that any conscious scene comprises a highly informative discrimination, and that is why they are “integrated” and “differentiated” at the same time. The main point of the dynamic core theory claims that conscious qualia are the discriminations mentioned above (Edelman, 2003). i.e. conscious processes " emerge " from neural processes, the latter "entail" of "give rise to" conscious properties by the "phenomenal transformation" that results in qualia; qualia "reflect" neural differentiations.

The hypothesis proposes that the neural mechanisms underlying consciousness start in the thalamocortical system. This system will be in charge of the re-entrance of neuronal interactions. The boundaries of the dynamic core are suggested to shift over time, when some neuronal groups or populations disappear and others are incorporated by the intrinsic and extrinsic adaptive characteristics that emerge in the course of an individual development, (Edelman and Tononi 2000).

Up to this point, we have seen several problems explaining the evolution of cognition with

30 an adaptational approach. The complexity of understanding this phenomenon includes replicator entities, structural genes and regulatory genes, developmental patterns, mutations, nervous system variations, neural groups, response strategies to external stimuli, and environmental interactions. All this, related to evolutionary processes like NS, neuronal group selection, and adaptation. Each of them are causal agents of cognition, a complexity of processes at different levels of organization. In the next subsection, I will develop the multilevel causal model for the evolution of cognition. Explaining in general terms what does multilevel causality tells of the levels of organization and the hierarchy ranges from molecules to ecosystems.

4.4. Multilevel causality

Biological systems can be understood as a hierarchical structure with different levels of organization that interact with each other form a network with various causal processes. For the arising of biological complexity, there will be numerous, quasi-independent modules at each level of the hierarchy, that will interact with each other in a network. A hierarchical structure describes a scale corresponding to its variables and to the description of the different levels of interaction of the same (Ellis 2009). For example, the complex chains of molecular interactions called metabolic pathways. As cell cycle pathways and developmental pathways, where cells grow, they divide and form more cells thanks to proteins that are the backbone of all biochemical pathways (Noble, 2006). In this way, cells are organized into tissues, such as skin, bones and muscles, to form organs such as the brain, heart and kidneys, and finally, all this together with the immune and hormonal systems, form the whole organism. This works in many different ways and at different levels of biological organization.

In the next section I will give a detail explanation of the types of multilevel causality to be able to recognize the most accurate one in the understanding of cognition.

4.4.1 Types of multilevel causality

We can distinguish two basic forms of multilevel causality: bottom-up causality and top-down causality. The first one according to Elllis (2006), is the capacity to reduce the levels of reality. Within biology, some scientists assume that there is a hierarchical organization of biological systems. Always seen as an increase in complexity from small to large. Thus, in the lower levels we have the molecules, proteins and cells; in the upper level are the tissues, organs, systems and individual. The lower levels have a causal power at higher levels, in some cases only the determination of what happens in them. In this way we know that in the cells there are proteins, that the tissues are formed by cells and so on. The core of this

31 reductionist view within science is strong, because all mechanisms can be explained bottom- up, based on the laws of physics and chemistry.

Top-down causality is the ability of higher levels of reality to have a causal power over lower levels (Ellis 2006). That is, altering the context of the higher level also alters the actions of the lower level. In such cases, the upper-level variables cannot be described in terms of the lower level, and this is what identifies them as context-dependent variables. An example is learning, when someone learns how to play the guitar, with practice his nervous system is reorganized and the fingers become more flexible so that it can carry out properly the action. Over time, this person becomes a 'virtuouso” musician. Of course, we can’t understand it’s 'virtuosity' from neural reorganization, as it would be an incomplete explanation. We must go to a higher level and resort to other causal agents such as beliefs and desires or social and cultural factors that would provide us with a more adequate and complete explanation. In the next sections we will apply the two types of causality models to cognition, focusing on the advantages and disadvantages of each model, in order to finally find a satisfactory model for the understanding of cognition.

4.4.1.1 Bottom-up causality for the understanding cognition.

The bottom-up causality is a strong element within the modern synthesis. The central dogma of biology describes that DNA is transcribed into RNA, that later on is translated into proteins, building the first hierarchy system in all live forms. The pyramid continues with the proteins that form the cells and so on. It is the equivalent of saying that the genotype creates the phenotype. It is with the same type of signals, that alterations will continue to affecting future brain responses and changes are reflected in the ability to repeat a physical or mental act. In this way, what the brain does is a constructive recategorization as experience is produced and not a precise replica of an earlier sequence of events. Global maps that relate to changing motion and sensory impressions are dynamic structures that are interconnected throughout the brain. They provide the substrate necessary to relate categorization to memory. Signals coming from the world or other parts of the brain trigger a selection process, which occurs at the level of synapses through an alteration. The altered synapses will depend on previous experience. As a consequence, there is a triggering between signals from neural groups that provoke a set of similar responses, those that in the past have had adaptive value. The property of neural circuit degeneration is what causes changes in memory as new experiences occur (Edelman and Tononi 2000).

32 In this way, the bottom-up recreation of memory would be explained in three causal levels. The first causal level is triggered by the sensory perception of an event from the outside world. The second level is a reconfiguration of neural activity based on the connections that were united by previous events with similar signals and that by their adaptive value are constantly reinforced. The third and last level is a selection process between the connections, in this way, some connections will be extended and others will be lost modifying the memory. In this way according to Edelman, memory in animals is always creative, rebuilt, and therefore not strictly replicative.

At the same time, cognitive abilities of animals also depend on properties of matter that are not specified in genes, such as epigenetic phenomena. So the need to be read through the phenotype rather than the reverse as dictated by the deterministic view (Astolfi et al., 2010). The success of explanations at the molecular level makes us forget that there is a process of feedback between genes and environment. Organisms also interact with the environment, and this will also have an impact on gene expression. Therefore, the functionality does not reside at the level of the genes. Genes are "blind" to what they do, which leads us to consider the difficulties of this bottom-up causality model.

One of the problems of bottom-up causality has to do with the fact that in every stage of life of an organism there are different patterns of gene expression. In the manufacture of a single protein, defining the chemical mechanisms involved in folding and specifying its activity is a very complex task. Simulations on the computer would take a lot of time. And the most important problem, even after a reliable reconstruction of the molecular level has been overcome, is that the structures and processes at the higher level are simply not visible at the molecular level. Genes and proteins do not reveal what is really happening at a higher level (Noble, 2009). Since the recipes that make up the mind and brain are always sensitive to the environment, there is no guarantee that they will converge to a concrete result. These recipes provide many different things, from the construction of enzymes and structural proteins, to the construction of engines, transporters, receptors and regulatory proteins, so there is no simple and easily characterized contribution of genetics to the mind (Marcus, 2003). It is pertinent to clarify that the difficulties presented by the bottom-up causality are basically due to the fact that it is an incomplete, not wrong view of the assumptions it handles. A hierarchical model of biological systems unidirectional, is not a satisfactory explanation to address some phenomenon. That is why it is necessary to resort top-down causation as we will see in the next section.

33 4.1.1.2 Top-down causality and cognition.

In this system, the higher levels influence the lower levels. As an example we can take a look into the interaction between nerve cells and the transmission of electrical signals. In all cells including neurons electrical potentials are given through the membrane. For a to operate, it is necessary for the electrical potential to change. To activate this change, an electric charge must be transmitted through the membrane. Molecules that carry electrical current through the membrane are called ions. Ions pass through membrane channels which in turn are proteins. For each protein molecule, there is at least one gene encoding it. One of the ions involved in this process is the sodium ion, which forms positive charges from the common salt or sodium chloride. The corresponding proteins, therefore, are called sodium channels. Then, you might think that to carry out electrical potentials that act quickly on a neuron, you have to express as many sodium channels as possible. However, in the 50's Alan Hodgkin, worked on the equations of electrical propagation in the nerves. He found that if the density of sodium channels is increasing, the electrical impulse is driven faster, but only to a certain extent. Once this point is reached, and the expression of the sodium channels continues, the transmission rate is reduced. Therefore, an effective way to maintain the smooth running process chain always moves in one direction. In an emergency sense: a higher unit can originate from the aggregation of lower units, but from the moment it acquires emerging properties by non-adaptive interaction between the parts (lower units), the upper unit becomes, by definition, an agent Independent in its own right, and not a passive 'slave' of the controlling constituents (Gould 2004), from genes to the entire organism. The clear trend is that one can appeal to reductionism. All phenomena can be explained by the lower elements genes and proteins, thus organisms are reconstructed from the bottom up. In this way we can talk about the existence of genes of the capacities, as the example of the FOXP2 gene.

In the case of neural Darwinism, we have the role of memory. This process is the result of a selective correspondence that occurs between the neural activity at that moment and the signals coming from the world. Signals first have to be categorized by the animal, and then associate this categorization with their previous experiences of nerves, to maintain a steady rate of expression of sodium channels, to the optimum level, but not beyond. This is how it works normally. When the system begins to become clogged with excess sodium, many channels stop being expressed. Thus, what happens in the upper level of the system influences the behavior of the lower level in the genes. The phenomenon called “electro- coupling”, it is a form of causality that operates from the top down and not from the bottom up. This applies to all genes that are expressed in the nervous system. To

34 change the frequency with which a nerve is excited, occurs during the synapse and the levels of gene expression change. Nerve cells feed back to their own nuclei and thus control the information of their own genes (Noble, 2006)

The above example is top-down causality, involving up to a group level of neurons. However, it can be given at higher levels, such as bottom-up causality through adaptive selection. Ellis (2009) tells us that this type of causality occurs when many entities interact. For example, the cells of a body or the individuals of a population. Variation occurs in the properties of these entities, followed by a selection of entities that are best suited to their environment or context. Higher-level environments provide niches that are favorable or unfavorable to particular types of lower-level entities. Variations that are best adapted to the niches are preserved and the others discarded. On this basis, the selection agent or an active element of the system accepts one of the states and rejects the rest. This state is selected and will constitute the starting point for the next round of selection which ultimately leads to the emergence and nature of the biological form. This generates new information that was not present before. This allows the emergence of complexity without specific goals that guide the process and the outcome is often not predictable.

Now, knowing how top-down causality operates through adaptive selection, we can relate neural Darwinism to this type of causality, based on the fact that allows the patterns of neural activity to adapt to new conditions. Sensory information from neurons is performed dynamically in order to adapt it to changes in environmental stimuli (Ellis, 2009). The key is that top-down causality occurs when the connections between neural groups are the basis of the upper-level adaptation criteria. These criteria, we recall, are provided by values that are responsible for guiding brain plasticity in response to the interactions mediated by learning in the environment and which is made effective by the neurotransmitters of the different brain maps. Adaptive selection is in operation when the selection of some of the varieties of neural groups that were formed from development, conserve some and discard the rest.

So far, we have just discussed the multilevel causality of cognition from an ontogenetic reconstruction. Nevertheless, it is also possible to see how it operates from the phylogenetic point of view. This point was developed by Martinez and Moya (2011) with its proposal of natural selection and top-down causality. They took as basis the arguments of Donald Campbell (1974) and mainly its fourth principle, the one of the descending causality. It asserts that nature is organized hierarchically at different levels by stating it as follows: when natural selection acts through life and death at a higher level of organization, the laws of the upper-level selective system determine, in part, the distribution of facts and substances of

35 the lower level (Campbell, 1983). Taking this principle, Martinez and Moya tell us that the downward causation of natural selection is possible thanks to the phenomenon of inheritance.

Thus, we can conclude that with more detailed studies of multilevel causality in the biological systems of cognition we can develop more complete explanations of evolutionary mechanisms that occur at a certain level and observe their consequences at other levels. We cannot just stay with the explanations that occur in a single level, under penalty of generating biased and restricted approaches. An explanation based strictly on evolutionary mechanisms at lower levels like genes, fails to explain what exists and occurs at higher levels. Likewise, we need to know what happens at the higher levels in order to explain the low levels. It is also important to consider that if a particular biological function or entity does not exist in a level, this does not mean that it does not exist at all. We will give a more detail explanation of this idea in the next section.

4.5 Integration of types of multilevel causality model to understand cognition

The evolution of cognitive abilities is the result of a network of complex interactions at different levels of organization. That is why throughout this work we have talked about cognition genes, brain development patterns, neural groups, and, learning behavior in the individual and across generations. This wide range of explanations can be explained with multilevel causality.

Multilevel causation can arise in different ways. It is important to always keep in mind both bottom up and top down causality. In this way we make sure we can have a reliable representation of the phenomenon. Multilevel causality can be expressed in an adaptational approach, as well as in an exaptationist7, ecological or social approach, with the same purpose, to represent the evolution of cognition. Causal relationships between different levels of biological organization are difficult to trace. That is why many times you can enter into conflict between different disciplines to come to represent the same phenomenon. The key to minimizing conflict is to always keep in mind that multilevel causality is a metaphor.

It is necessary to take multilevel causality as a metaphor not because one doubts the ontological status of biological entities: genes, proteins, cells, organs, etc. But simply the dynamism of epistemology at each level makes it impossible to take it as real. When I draw a

7 This term refers to a process in which a feature acquires a function that was not acquired through NS. Was given by Gould in 1982

36 causal line between genotype and cerebral plasticity, I know that I am ignoring many phenomena that occur and that I am not taking into account. In the same way, when a causal line between gene and protein is drawn, although it is occurring at very close organizational levels, we know that there are intermediaries between these two that are not taken into account. Not because its importance is minimal or want to deny its existence. But a multilevel causal model also reflects a synthesis of biological complexity.

37 Conclusions

Cognition is a complex process. Since the definition to the study of it, cognition bring a lot of questions to the table. We examined, that in order to study the adaptationist approach of cognition, it is important to consider cognition as an historical process. The approach given by Dennett, takes into account this consideration. That is, many animals including humans have the same cognitive abilities as perception, memory and learning, among others, because there is an evolutionary continuity due to a common ancestor. However, reviewing Dennett’s approach, we realized that his theory had a lot of blank spaces in between. In this case, what is not clear is why all cognitive abilities evolved as adaptations product of natural selection. For example, he claims that learning a 'good trick' evolved by natural selection. On the contrary, it is known that not all the 'good tricks' that animals learn throughout their lives are functional for their survival and reproduction, because the next generation might not be part of the same ecosystem. Our intention is not to deny that 'some' cognitive abilities are a product of selection, but that Dennett is not able to define which ones are and which are not.

In this work, we found Dennett’s evolutionary argument of cognition excessively speculative. This is due to the presentation of natural selection as the only force responsible of the adaptation of cognitive abilities. He does not resort to any evolutionary evidence such as structural and/or behavioral analogies or homologies to support his account. He simply assumes that in each generation the individuals who 'better' perceive, 'better' learn, and so on, survive and reproduce without presenting the circumstances in which each selection process was carried out.

For its part, neural Darwinism proposes that natural selection acts at the level of nerve cells. Taking the arguments of Gould (2004), and in agreement with him, the neurons would have to fulfill three criteria that interact together so that they can be considered units of Darwinian selection (Darwinian individuality). These criteria are: reproduction, variation and inheritance. And as we saw in Chapter 2, neural Darwinism does not meet inheritance. Therefore it is concluded that natural selection can not be acting at the level of neural groups.

Dennett's Darwinist arguments and Edelman's neural Darwinism are just two examples of what Rose (2001) called 'ultra-Darwinism'. In agreement with Rose, we consider that both authors present Darwinian evolution as a dogma competent to apply to any cognitive phenomenon. As a consequence the theories of both authors are incomplete or erroneous as we have study.

38

Due to the partial approaches offered by Dennett and Edelman to explain the evolution of cognition. It is necessary to appeal to other arguments that could reveal more about the origin and persistence of some cognitive capacities. Such is the case of the arguments of Evo-Devo. Nervous systems of animals during development are subject to physical or chemical constraints and also to historical contexts. Limiting the variation of the nervous systems. The origin of the variations of cognitive-cerebral structures would not only be the result of mutations as established by the modern synthesis. Each species has its own trajectory of brain development, in that period development occurs processes that are not dictated exclusively by the genes. In this way it could be explained the different types of nervous systems that come from ancestors in common, as well as the possible patterns of behavior that they can generate, without resorting to only an adaptive explanation.

When we consider organisms as open systems, we leave aside the view of passive agents at the expense of the environment. Cognitive abilities not only give individuals tools to solve environmental problems, but also to create their own environments. Seen in this way, the organism by means of cognition creates its own spaces, modifies them and inherits them generation after generation, sometimes the instructions for this to happen surpass genes or development processes. So, we can considerate that the evolution of cognitive abilities is the result of a network of complex interactions at different levels of organization. That is why throughout this work we have talked about cognition genes, brain development patterns, neural groups, and, learning behavior in the individual and across generations. With such a wide range of explanations is when we can resort to multilevel causality.

39 References • Emery, N. J., & Clayton, N. S. (2004). The mentality of crows: convergent evolution of • Anderson, J. R. (2013). The architecture of intelligence in corvids and cognition. Psychology Press. apes. science, 306(5703), 1903-1907. • Astolfi, P. A., Salamini, F., & Sgaramella, • evolution, 28(12), 712-718. V. (2010). Are we genomic mosaics? • Fiddick, L., & Barrett, H. C. (2001). Variations of the genome of somatic cells Evolution of cognition. can contribute to diversify our • Gehring, W. J. (1992). The homeobox in phenotypes. Current genomics, 11(6), 379- perspective. Trends in biochemical 386. sciences, 17(8), 277-280. • Balari, S., & Lorenzo, G. (2009). • Gilbert, S. F. (2000). Developmental Computational phenotypes: Where the biology, Sinauer Associates theory of computation meets Evo- • Godfrey-Smith, P. (2002). Environmental Devo. Biolinguistics, 3(1), 002-060. complexity and the evolution of cognition. T • Bateson, P., & Laland, K. N. (2013). he evolution of intelligence, 233-249. Tinbergen's four questions: an appreciation • Gould, S. J. (1977). Ontogeny and and an update. Trends in ecology & phylogeny. Harvard University Press. • Carroll, S. B. (2008). Evo-devo and an • Gould, S. J. (1977). Ontogeny and expanding evolutionary synthesis: a phylogeny. Harvard University Press. genetic theory of morphological evolution. • Gray P.H. 1967. Spalding and his influence Cell, 134(1), 25-36. on research in developmental behavior. • Churchland, P. M. (1989). A Journal of the History of the Behavioral neurocomputational perspective∫ The • Hamilton, W. D. (1964). The genetical nature of mind and the structure of science. evolution of social behaviour. I. J ournal of MIT press. theoretical biology, 7(1), 1-16. • Churchland, P. M. (2013). Matter and • Hamilton, W. D. (1964). The genetical consciousness. MIT press. evolution of social behaviour. II. J ournal of • Darwin, C. (1871) The Descent of Man and theoretical biology, 7(1), 17-52. collection in relation to sexÆ London,John • Kornblith, H. (2007). The Metaphysical Murray. Status of Knowledge1. P hilosophical • Darwin, C. (1872) The expression of Issues, 17(1), 145-164. emotions in man and animalsÆ • Lamarck, J. D. (1809). Philosophie London,John Murray. zoologique. English version (1914): • Dennett, D. (2001). Are we explaining Zoological philosophy. consciousness yet?. Cognition, 79(1), 221- • Lashley, K. S. (1929). Brain mechanisms 237. and intelligence: A quantitative study of • Dennett, D. C. (1971). Intentional systems. injuries to the brain. The Journal of Philosophy, 68(4), 87-106. • Lashley, K. S. (1950). In search of the • Dennett, D. C. (1991). Consciousness engram. explained. Little, Brown.[SB](1995) • Lashley, K. S. (1951). The problem of Darwins dangerous idea. Penguin/Simon & serial order in behavior. In C erebral Schuster.[SB, DD, aAM](2001) The mechanisms in behavior (pp. 112-136). evolution of culture. The • Lewontin, R. C. (1998). The evolution of Monist, 84(30524), E83E92. cognition: Questions we will never answer. • Diéguez Lucena, A. (2011). La evolución An invitation to cognitive science, 4, 107- del conocimiento. D e la mente animal a la 132. mente humana, 14, 00. • Lewontin, R. C. (2001). The triple helix: • Donald, M. (1991). Origins of the modern Gene, organism, and environment. Harvard mind∫ Three stages in the evolution of University Press. culture and cognition. Harvard University • Marcus G. (2005) El nacimiento de la • Edelman, G. M. (1987). Neural Darwinism: mente. Como un numero pequeñísimo de The theory of neuronal group selection. genes crea las complejidades del Basic Books. pensamiento humano. Barcelona. Ariel • Edelman, G. M. (1993). Neural Darwinism: • Neander, K. (2007). Biological approaches selection and reentrant signaling in higher to mental representation. P hilosophy of brain function. Neuron, 10(2), 115-125. Biology¨ Amsterdam∫ Elsevier, 549-565. • Edelman, G. M. (1994). The evolution of • Nicholls, J. G., Martin, A. R., Wallace, B. somatic selection: the antibody G., Brown, D.A., Diamond M.E. & Fuchs, tale. Genetics, 138(4), 975. P. A. (2012). From neuron to brain. • Edelman, G. M., & Tononi, G. (2000). A Sunderland, MA: Sinauer Associates. universe of consciousness: How matter • Papini, M. R. (2009). Psicología becomes imagination. Basic books. comparada∫ evolución y desarrollo del • Edelman, G. M., & Tononi, G. comportamiento. Editorial El Manual (2013). Consciousness: How matter Moderno Colombia. becomes imagination. Penguin UK.

40 • Parker, S. T., & McKinney, M. L. intelligence, 1882. Mental Evolution in (2012). Origins of intelligence: The Animals. evolution of cognitive development in • Sciences. 3 , 168-179. monkeys, apes, and humans. JHU Press. • Spalding D.A. 1872. On instinct. Nature, 6, • Peláez del Hierro, F., Gil, C., & Sánchez, 485-486. S. (2002). Introducción a la etología. • Thorpe W.H. 1979. The origins and rise of Madrid Editorial Biblioteca Nueva. ethology the science of the natural • Pocheville, A. (2015). The ecological niche: behavior of animals. Heinneman, London, history and recent controversies. 18-27. In Handbook of evolutionary thinking in the • Van Horik, J., & Emery, N. J. (2011). sciences (pp. 547-586). Springer Evolution of cognition. Wiley Netherlands. Interdisciplinary Reviews Cognitive • Press. Science, 2(6), 621-633. • Roitblat, H. L., & Meyer, J. A. (1995). • Wilson, E. (1978). 0.(1975) Sociobiology: Comparative approaches to cognitive The New Synthesis. science. Mit Press. • Romanes, G. J. (1883). Animal

41