DynaPsych Table of Contents


 

 

 

Biological versus computational approaches to creative problem solving

 

Robert Arp

Department of Philosophy

Southwest Minnesota State University

1501 State Street

Marshall, MN 56258

arpr@southwestmsu.edu

 

Routine problem solving versus creative, nonroutine problem solving

Imagine being a dentist in the early part of the 19th century.  Now, imagine going to the dentist to have a tooth pulled in the early part of the 19th century.  In those days, pulling teeth was a painful experience for the patient, as there were no known anesthetics being used at the time.  The kinds of things a dentist used to help ease the patient’s pain before a tooth extraction might have included having the patient suck on a medicinal herb that produces a numbing effect in the mouth, placing ice upon the gums, getting the patient to drink alcohol before the procedure, or any combination thereof.  Such were the methods that Dr. Horace Wells likely used to solve the problem of pain associated with tooth extractions while working as a dentist in Hartford, Connecticut circa 1844.  These methods probably were nothing new, and we can imagine that dentists had been using these remedies for some time so as to alleviate or prevent the pain associated with tooth extraction.

One evening in 1844, Dr. Wells attended an amusing public demonstration of the effects of inhaling a gas called nitrous oxide with his friend, Samuel Cooley.  Mr. Cooley volunteered to go up on stage to inhale the gas, and proceeded to do things like sing, laugh, and fight with the other volunteers who had inhaled the gas.  In the scuffle, Cooley received a deep cut in his leg before sobering up and coming back to his seat next to Dr. Wells.  Someone noticed a pool of blood under Cooley’s seat, and it was discovered that Cooley had cut his leg.  However, Cooley seemed to be unaffected by and was unaware of the wound.  Upon witnessing this event, a light went on in Dr. Wells’ head: What if this laughing gas could be used during tooth extraction to ease a patient’s pain?  The problem of pain associated with tooth extraction finally might be solved!  In fact, over the next several years Dr. Wells proceeded to use nitrous oxide, was successful at painlessly extracting teeth from his patients, and the seeds of modern anesthesia were sewn (Roberts, 1989).

Humans are resourceful animals.  We can imagine Dr. Wells prescribing the various remedies of which he was familiar – the medicinal herb, the ice, the alcohol – in the attempt to ease a patient’s pain during tooth extraction.  In such a case, we would have an instance of what Mayer (1995) has called routine problem solving, whereby a person recognizes many possible solutions to a problem given that the problem was solved through one of those solutions in the past.  People constantly perform routine problem solving activities that are concrete and basic to their survival, equipping them with a variety of ways to “skin” the proverbial cat, as well as enabling them to adapt to situations and re-use information in similar environments.

However, we also can engage in activities that are more abstract and creative such as invent tools based upon mental blueprints, synthesize concepts that, at first glance, seemed wholly disparate or unrelated, and devise novel solutions to problems.  When Dr. Wells decided to use nitrous oxide on his patients, he pursued a wholly new way to solve the problem of pain.  This was an instance of what Mayer (1995) has called nonroutine creative problem solving.  Nonroutine creative problem solving involves finding a solution to a problem that has not been solved previously.  The introduction of nitrous oxide in order to extract teeth painlessly would be an example of nonroutine creative problem solving because Dr. Wells did not possess a way to solve the problem already, and he had not pursued such a route in the past.

Not only do people make insightful connections like that of Dr. Wells, they take advantage of serendipitous opportunities, invent products, manufacture space shuttles, successfully negotiate environments, hypothesize, thrive, flourish, and dominate the planet by coming up with wholly novel solutions to problems.  It is possible to speak about creative problem solving from the biological, evolutionary perspective as well as from a non-biological, computational perspective.  After all, researchers have been able to get both biological entities and alife creations – those that are purely simulated, as well as those that are connected to robotic mechanisms – to engage in fairly complex forms of problem solving.  In this paper, after putting forward a view of problem solving from the biological/evolutionary perspective that is rooted in Mithen’s (1996, 1999) idea of cognitive fluidity, I go on to argue that nonroutine creative problem solving – as opposed to routine problem solving – most likely is only possible for biologically conscious entities having an epigenetic history.

On the negative side, I argue that the dry-mind or computational approach to creative problem solving in biologically conscious minds is at the very least deficient and, at best, should be combined with wet-mind biological and evolutionary approaches.  I do this by showing that the neural circuitry necessary for creative problem solving in humans evolved for the purposes of negotiating specific organic environments and, in this sense, is fundamentally of a different kind from that of computational circuitry.  I also present evidence of the failure of computational models to handle even routine forms of problem solving, let alone nonroutine forms of creative problem solving.

On the positive side, I acknowledge that computational models of simulated (or virtual) mental activity, set up to “evolve” in simulated environments, are useful to our understanding of how real minds function in the real world.  Thinkers doing work in cognitive science and artificial intelligence have made important contributions to biology and neuroscience, and there is a mutual benefit to be gained – in terms of our understanding of the mind, its activities, and its evolution – by these dry-mind and wet-mind groups working together.

Bissociation and Mithen’s idea of cognitive fluidity

It is important to elaborate further upon the distinction between routine problem solving and nonroutine creative problem solving.  From the previous section, we already know that routine problem solving deals with the recognition of many possible solutions to a problem, given that the problem was solved through one of those solutions in the past.  Here, we can link routine problem solving to the kind of trial-and-error strategizing and calculation that animals other than human beings typically engage in, although humans engage in routine problem solving as well.  In this sense, routine problem solving entails a mental activity that is stereotyped and wholly lacking in innovation, because there are simply perceptual associative connections being made by the mind of an animal.  Images in perception or memory are associated with one another and/or with some environmental stimuli so as to learn some behavior, or produce some desired result.  If that result is not achieved, an alternate route is pursued in a trial-and-error fashion.

For example, Olton & Samuelson (1976) showed that rats are able to associate routes in a maze with food acquisition.  In these experiments, food was placed at the end of each arm of an 8-arm radial maze, and a rat was placed in the center of the maze and was kept in the maze until all the food was collected.  At first, the rat did not associate a certain path with the food.  But after trial-and-error, the rat eventually got all of the food.  In subsequent tests, the food was placed in the same spot in the maze, and the same rat was able to more quickly and efficiently associate the correct pathway with the acquisition of food.

Associative learning tests have been performed on humans and animals numerous times (Maier, 1932; Zentall et al., 1990; Dickinson, 1980; Rescorla, 1988; Macphail, 1996, 1998; Mackintosh, 1983, 1995; Hall, 1994, 1996).  In his famous delayed matching to sample tests, Hunter (1913) demonstrated that rats, raccoons, and dogs are able to associate memories of a stimulus with the same stimulus perceived by the animal, so as to solve some problem.  Wright (1989, 1997) has shown that pigeons and monkeys can perform similar associations (also Roberts & Grant, 1974).  A typical battery of I.Q. tests will have several association tests whereby people are asked to solve routine problems, such as linking a word to a picture and/or linking pictures to one another in a familiar sequence (Sternberg, 1996, 2000, 2001).

Concerning nonroutine creative problem solving, we already know that this entails pursuing a wholly new way to solve a problem that has not been solved previously, and that the problem-solver did not possess a way to solve the problem already.  Here, however, we can draw a distinction between solving a nonroutine problem through imitation with another’s help, and solving a nonroutine problem on one’s own.  Some animals appear to have the capacity to solve nonroutine problems, once the solutions have been shown to them, or imitated for them.

Consider the following cases that demonstrate an animal’s ability to creatively problem solve through imitation with another’s help.  An octopus studied by Fiorito et al. (1990) has been documented as being able to unpop the cork on a jar to get at food inside.  Initially, the octopus could see the food in the jar, but was unable to unpop the cork of the jar to get at the food.  The next time, Fiorito et al. unpopped the cork while the octopus was watching, resealed the jar, and gave it to him in his tank.  The octopus was able to associate the unpopping of the cork with the acquisition of food, apparently remembered what Fiorito et al. had shown him, and unpopped the cork himself to get at the food.

Also, we have documented chimps trying a couple of different ways to get at fruit in a tree – like jumping at it from different angles or jumping at it off of tree limbs – before finally using a stick to knock it down.  Scientists also document young chimps watching older chimps do the same thing (Tomasello, 1990; Tomasello et al., 1987, 1993; Byrne, 1995; Savage-Rumbauch & Boysen, 1978; Whiten et al., 1996, 1999).  Like the octopus’ problem solving ability, this seems to be a form of nonroutine creative problem solving by use of another’s help.

In fact, several observations have been made of various kinds of animals engaged in imitative behaviors: Whiten & Custance (1996), Whiten et al. (1996), Tomasello et al. (1993), and Abravanel (1991) has documented imitative behaviors in chimpanzees and children; Parker (1996), Miles, Mitchell, & Harper (1996), Call & Tomasello (1994), and Russon & Galdikas (1993, 1995) have witnessed young orangutans imitating older orangutans using sticks and rocks to gather food, as well as throw sticks and rocks at other orangutans in self-defense; Yando, Seitz, & Ziqler (1978), Mitchell (1987), and Moore (1992) report mimicry and imitation in birds; and Heyes & Dawson (1990) and Heyes, Jaldon, & Dawson (1992) note evidence of imitative behaviors in rats.

However, the number of possible solution routes is limited in these examples of routine problem solving.  If either the octopus’ corked jar was sealed with Crazy Glue, or there were no sticks around, or there were no other older chimps or researchers around to show younger chimps how to use sticks, the octopus and chimpanzees in the above cases likely would starve to death.  The possible solution routes are limited because the mental repertoire of these animals is environmentally fixed, and their tool usage (if they have this capacity) is limited to stereotypical kinds of associations.

Bitterman (1965, 1975) tested the intelligence levels of fish, turtles, pigeons, rats, and monkeys with a variety of tasks, including pushing paddles in water, pecking or pressing lighted disks, and crawling down narrow runways.  Although such animals improved their abilities to perform these tasks as time went on, Bitterman found that these species only could perform a limited number of associative learning tasks.  This data, along with the data concerning the octopus, chimps, orangutans, rats, and birds supports the idea that these animals are engaged in mostly habitual, stereotyped forms of associative thinking and learning (cf. the new research concerning crows and other birds in Weir, Chappell, & Kacelnik, 2002; Emery & Clayton, 2004; Reiner, Perkel, Mello, & Jarvis, 2004).

Unlike routine problem solving, which deals with associative connections within familiar perspectives, nonroutine creative problem solving entails an innovative ability to make connections between wholly unrelated perspectives or ideas.  Again, this kind of problem solving can occur as a result of imitation through another’s help – as in the above octopus and chimpanzee examples – as well as on one’s own.  A human seems to be the only kind of being who can solve nonroutine problems on his/her own, without imitation or help.  This is not to say that humans do not engage in solving nonroutine problems through imitation; in fact, nonroutine creative problem solving by imitation occurs all of the time, especially in the earlier years of a human’s life.  This is just to say that humans are the only animals who have the potential to consider wholly new routes to problem solving.

Koestler (1964) referred to this quality of the creative mind as a bissociation of matrices.  When a human bissociates, that person puts together ideas, memories, representations, stimuli, and the like in wholly new and unfamiliar ways for that person.  Echoing Koestler, Boden (1990) calls this an ability to “juxtapose formerly unrelated ideas” (p. 5; see Terzis 2001).  Thus, Dominowski (1995) claims that “overcoming convention and generating a new understanding of a situation is considered to be an important component of creativity” (p. 77).

When animals associate, they put together perceptions, memories, representations, stimuli, and the like in familiar ways.  For example, my cat associates my loud voice with her being in trouble (and runs away), the rat associates a route with food, and the octopus associates corked-jar-experiment-B with corked-jar-experiment-A and more quickly can unpop the cork on the jar to get at the food in subsequent tests.  As far as we know, animals can associate only, so they always go for solutions to problems that are related to the environment or situation in which they typically reside.  Humans bissociate, and are able to ignore normal associations, and try out novel ideas and approaches in solving problems.  Such an ability to bissociate accounts for more advanced forms of problem solving whereby the routine or habitual associations are the kinds of associations that precisely need to be avoided, ignored or bracketed out as irrelevant to the optional solution (Finke, Ward, & Smith, 1992).  Bissociation also has been utilized in accounting for risibility, hypothesis-formation, art, technological advances, and the proverbial “ah-hah,” creative insight, eureka moments humans experience when they come up with a new idea, insight, or tool (Koestler, 1964; Boden, 1990; Holyoak & Thagard, 1995; Terzis, 2001; Davidson, 1995).

So, when we ask how it is that humans can be creative, part of what we are asking is how they bissociate, viz., juxtapose formerly unrelated ideas in wholly new and unfamiliar ways for that person.  To put it crudely, humans can take some idea “found way over here in the left field of the mind” and make some coherent connection with some other idea “found way over here in the right field of the mind.”  Humans seem to be the only species that can engage in this kind of mental activity.  How is this possible?

Steven Mithen (1996, 1999) has put forward the notion of cognitive fluidity, an idea that serves the purpose of enabling one to respond creatively to nonroutine problems in environments.  Mithen’s idea has merit because, as he notes, he is an archeologist who is applying the hard evidence of evolutionary theory, fossils, and toolmaking to psychology.  Not only is he speculating about the mind, but he has the archeological evidence to support his speculations.  Fodor (1998), Calvin (2004), and Stringer & Andrews (2005) praise Mithen’s idea of cognitive fluidity as being a significant hypothesis, as well as consistent with archeological and neurobiological evidence.  As a philosopher of mind and biology, I applaud Mithen’s hypothesis as well (Arp, 2005a, 2005b, 2006a, 2006b, 2007a, 2007b).

Mithen (1996) sees the evolving mind as going through a three-step process.  The first step begins prior to 6 mya when the primate mind was dominated by what he calls a general intelligence.  This general intelligence consisted of an all-purpose, trial-and-error learning mechanism that was devoted to multiple tasks.  All behaviors were imitated, associative learning was slow, and there were frequent errors made, much like the mind of the chimpanzee.

The second step coincides with the evolution of the Australopithecine line, and continues all the way through the Homo lineage to H. neandertalensis.  In this second step, multiple specialized intelligences – or modules, as evolutionary psychologists call them – emerge alongside general intelligence.  Associative learning within these modules was faster, and more complex activities could be performed.  Compiling data from fossilized skulls, tools, foods, and habitats, Mithen concludes that H. habilis probably had a general intelligence, as well as modules devoted to social intelligence (because they lived in groups), natural history intelligence (because they lived off of the land), and technical intelligence (because they made tools).  Neandertals and H. heidelbergensis would have had all of these modules, including a primitive language module, because their skulls exhibit bigger frontal and temporal areas.  According to Mithen, the neandertals and members of H. heidelbergensis would have had the Swiss Army knife mind that evolutionary psychologists speak about (Cosmides & Tooby, 1987, 1992, 1994; Pinker, 1994, 1997, 2002; Shettleworth, 2000; Gardner, 1993; Scher & Rauscher, 2003; Plotkin, 1997; Palmer & Palmer, 2002).

At this point, we note a criticism Mithen makes about the evolutionary psychologists who think that the essential ingredients of mind evolved during the Pleistocene epoch.  It concerns the simple fact that modern-day humans deal with a whole different set of problems to overcome than did our Pleistocene ancestors.  We can look back to the environment of the Pleistocene and note how certain cognitive features emerge, and become part of, the normal genetic makeup of the human.  However, as Mithen (1996) asks: “How do we account for those things that the modern mind is very good at doing, but which we can be confident that Stone Age hunter-gatherers never attempted, such as reading books and developing cures for cancer” (pp. 45-46)?

The emergence of distinct mental modules during the Pleistocene that evolutionary psychologists like Cosmides, Tooby, and Pinker speak about as being adequate to account for learning, negotiating, and problem solving in our world today cannot be correct.  For Mithen, the potential variety of problems encountered in generations subsequent to the Pleistocene is too vast for a much more limited Swiss Army knife mental repertoire; there are just too many situations for which nonroutine creative problem solving would have been needed in order to not only simply survive, but also to flourish and dominate the earth.  Pinker (2002) thinks that there are upwards of 15 different domains, and various other evolutionary psychologists have their chosen number of mental domains (e.g., Buss, 1999; Shettleworth, 2000; Gardner, 1993; Plotkin, 1997; Palmer & Palmer, 2002).  However, there are potentially an infinite number of problems to be faced on a regular basis by animals as they negotiate environments.  It does not seem that there would be a way for 15, 20, 25 – or even 1,000 – domains to handle all of these potential problems (Buller, 2005).  That we negotiate environments so well shows that we have some capacity to handle the various and sundry potential nonroutine problems that arise in our environments.

Here is where the third step in Mithen’s (1996) evolution of the mind comes into play known as cognitive fluidity.  In this final step, which coincides with the emergence of modern humans, the various mental modules are working together with a fluid flow of knowledge and ideas between and among them.  The information and learning from the modules can now influence one another, resulting in an almost limitless capacity for imagination, learning, and problem solving.  The working together of the various mental modules as a result of this cognitive fluidity is consciousness for Mithen, and represents the most advanced form of mental activity.

Mithen uses the schematization of the construction of a medieval cathedral as an analogy to the mind and consciousness.  Each side chapel represents a mental module.  The side chapels are closed off to one another during construction, but allow people to have access from the outside to attend liturgies, much like mental modules are closed off to one another (encapsulated) and have specified input cues.  Once the cathedral chapels have been constructed and the central domed superchapel is in place, the doors of all of the chapels are opened, and people are allowed to roam freely from chapel to chapel.  Analogously, modern humans have evolved the ability to allow information to be freely transmitted between and among mental modules, and this cognitive fluidity comprises consciousness.

Mithen goes on to note that his model of cognitive fluidity accounts for human creativity in terms of problem solving, art, ingenuity, religion, and technology.  His idea has initial plausibility, since it is arguable that the neandertals died off because they did not have the conscious ability to re-adapt to the changing environment.  It is also arguable that humans would not exist today if they did not evolve consciousness to deal with novelty (Arp, 2005a, 2005b, 2006b, 2007a; Bogdan, 1994; Cosmides & Tooby, 1992; Gardner, 1993; Humphrey, 1992; Pinker, 1997).  No wonder, then, that Crick (1994) maintains: “without consciousness, you can deal only with familiar, rather routine situations or respond to very limited information in new situations” (p. 20).  Also, as Searle (1992) observes: “one of the evolutionary advantages conferred on us by consciousness is the much greater flexibility, sensitivity, and creativity we derive from being conscious” (p. 109).  Modular processes can be used to explain how the mind functions in relation to routinely encountered features of environments (Jackendoff, 1987; Dennett, 1991; Hale, 1989; Kimura, 1989; Crick & Koch, 1999).  However, depending on the radicalness of a novel environmental feature, inter-modular processes (Mithen’s cognitive fluidity) may be required to deal effectively and, at times, creatively with the problem.

At this point, we must speak about the importance of the effect that novel environments have on the brain.  There is now solid evidence that the environment contributes to the formation, maintenance – even re-growth or co-opting – of neurons and neural processes in the brain.  For example, it has been shown that neuronal size and complexity, as well as numbers of glial cells, increase in the cerebral cortices of animals exposed to so-called enriched environments, viz., environments where there are large cages and a variety of different objects that arouse curiosity and stimulate exploratory activity (Diamond, 1988; Diamond & Hopson, 1989; Mattson, Sorensen, Zimmer, & Johansson, 1997; Receveur & Vossen, 1998).  Also, there is recent data suggesting that regions of the brain can be trained, through mental and physical exercises, to pick up tasks from other regions (Holloway, 2003; van Praag, Kempermann, & Gage, 2000; Schwartz & Begley, 2002; Clayton & Krebs, 1994).

The implications of synapse strengthening from environmental stimuli, as well as the ability of neuronal processes to perform alternate functions, are integral to an evolutionary explanation of conscious creative problem solving in humans.  This so because the novel promotes unusual or extreme stimulation of cells, such stimulation of cells causes new connections to be made in the brain, new connections cause better response of the animal to external stimuli, and better response causes likelihood of survival so as to pass genes on to progeny.

Again, environment is only half of the two-sided biological coin that includes nurture (the environmental influence) as well as nature (the genetic influence).  On the genetic side, chance mutations cause a trait – like the neocortex and consciousness that emerges from it – to come to be, this trait may be useful in some environment, the animal with that trait may survive to pass it on to its progeny, and this is an endless progressing cycle of genetic adjustment, re-adjustment, adjustment, re-adjustment, etc.  It is wholly plausible that the mental properties necessary for creative problem solving evolved from this interplay of genes and a novel environment.  Thus, Barlow (1994) maintains: “Anything that improves the appropriateness and speed of learning must have immense competitive advantage, and the main point of this proposal is that it would explain the enormous selective advantage of the neocortex.  Such an advantage, together with the appropriate genetic variability, could in turn account for its rapid evolution and the subsequent growth of our species to its dominant position in the world” (p. 10).  This aforementioned information is significant to Mithen’s account of cognitive fluidity because, given the novelty our early hominins dealt with in their environments, we can see how it would have been possible for newer connections between areas of the brain to have been made, as well as how wholly new connections could have arisen, acting as the neurobiological conditions for cognitive fluidity.

Mithen’s idea of cognitive fluidity helps to explain our ability to bissociate because the potential is always there to make innovative, previously unrelated connections between ideas or perceptions, given that the information between and among modules has the capacity to be mixed together, or intermingle.  So in essence, cognitive fluidity accounts for bissociation, which accounts for human creativity in terms of problem solving, art, ingenuity, and technology.  This is not to say that the information will in fact mix together, and then be bissociated by an individual.  This is just to say that there is always the potential for such a mental process to occur in our species.  In the words of Finke et al. (1992): “people can generate original images that lead to insight and innovation or commonplace images that lead nowhere, depending on the properties of those images” (p. 2).

Dry mind versus wet mind

Now, it is possible to speak about creative problem solving from the biological, evolutionary perspective as well as from a non-biological, computational perspective.  After all, researchers have been able to get both biological entities and alife creations – those that are purely simulated, as well as those that are connected to robotic mechanisms – to engage in fairly complex forms of problem solving.  The dry-mind, computational approach to problem solving begins with the idea that the processes and procedures that computers go through are analogous to the mind’s processes and procedures.  Since the 1950s, computers have been utilized as a model for the workings of the mind (Turing, 1950; von Neumann, 1958; Jackendoff, 1987; Holyoak & Thagard, 1995; Copeland, 1993).  Such an idea makes sense, since it seems that both the mind and a computer do things like calculate, compute, and engage in algorithmic processes, as well as solve problems.  Also, such a close connection between computers and minds makes sense because, after all, it is human minds that program computers to go through computations, calculations, algorithms, and the like!

Some thinkers have become so enamored by the close affinity between computational processes and mental processes that they think the mind just is a computational process.  This is called the strong artificial intelligence position.  Those thinkers who share this fundamental metaphysical position will pursue methodological routes whereby they set up computer programs and robotic mechanisms, run various tests using these programs and mechanisms, and then utilize the results to make direct inferences about the workings of the mind.  By this thinking, then, one day it will be possible to construct computers, robots, androids, synthetics, or cyborgs that will not only behave like human beings, but also will have the same mental states as human beings, including consciousness.  Kosslyn & Koenig (1995) call this the dry-mind approach to understanding the mind, in contradistinction to the wet-mind approach.

The dry-mind approach to understanding the mind is called dry because the investigation of the mind takes place at the computational level, wholly removed from the hardware of the brain on which such mental activity takes place, or from which such mental activity emerges.  Consider Pylyshyn’s (1980) dry-mind claim that “in studying computation it is possible, and in certain respects essential, to factor apart the nature of the symbolic process from the properties of the physical device in which it is realized” (p. 115).  Conversely, if one studies the processes and systems of the brain itself to understand the workings of the mind, then one is dealing with the moist piece of matter that is the substrate for such computational types of processing.  Given that the brain is a moist piece of matter, Kosslyn & Koenig (1995) call this methodology wet.

In his influential work on vision, Marr (1983) described three levels of analysis of perceptual information processing.  The first is a computational level, whereby we seek the information processing goal or task.  The second level deals with how the information is represented, as well as the algorithm for transforming it.  The third level deals with how the algorithm is implemented in the hardware.  We can note two important implications of this tripartite system for the distinction between dry-mind and wet-mind approaches.

First, the first two levels of Marr’s system are consistent with the position in the philosophy of mind and cognitive science known as computational functionalism.  Computational functionalists generally are advocates of strong artificial intelligence, and they see the mind in terms of the causal/functional relations between inputs, outputs, and other mental (i.e., functional) states of some system.  The physical realization in some system is not the essence of the mind; rather, the mind is characterized in terms of its role in relating inputs to outputs, and its relation to other functional components of the system (see Block, 1994; Fodor, 1983, 1998; Putnam, 1960; Churchland, 1986).

Second, the third level of Marr’s system deals with how algorithms can be implemented in the hardware.  Note the usage of the word hardware.  According to functionalists who advocate strong artificial intelligence, it is possible for intelligent (software) behavior to be realized or emerge from any potential material medium (hardware).  If silicon and metal in a robot or some plasma in an alien is set up to function in the right way, then such robots or aliens will have mental states.  Further, depending upon the sophistication of the causal connections, such robots or aliens would have conscious mental states.  The important point is that it makes no difference to the functionalist what material medium is utilized, as long as that material medium is functioning in the right way so as to yield intelligent behavior (Block, 1980; Churchland, 1986; Mahner & Bunge, 2001).

A basic approach computational functionalists utilize in trying to describe the mind is known as connectionism.  According to advocates of connectionism, mind can be described as a vast network of nodes, which are supposed to be analogous to neurons, and whose different and variable excitation levels explain mental activity (Feldman & Ballard, 1982; Bechtel & Abrahamsen, 2002; Smolensky, 1988; Rumelhart & McClelland, 1985).  Interestingly enough, connectionist models are structured on the concept of the parallel processing involved in neural networks.  Such a pattern consists of units, or nodes, that are linked together by numerous pathways.  Like the tripartite neuronal layout of the brain, in a connectionist model there are generally three classes of nodes, viz., input nodes, hidden nodes, and output nodes.  Input nodes receive information from the world, output nodes show the result of processing, and the hidden nodes in between carry out the processing.  Nodes receive signals from other nodes, and output signals to other nodes, in this huge interconnected parallel processing network.  The mind, then, just is the entire networking process.  Dennett (1991) endorses this view, and he notes that the consciousness mind “can be best understood as the operation of a ‘von Neumannesque’ virtual machine implemented in the parallel architecture of a brain” (see von Neumann, 1958).  Dennett’s justification for this position is simply this: “Since any computing machine at all can be imitated by a virtual machine on a von Neumann machine, it follows that if the brain is a massive parallel processing machine, it too can be perfectly imitated by a von Neumann machine” (p. 217).

Now, such advocates of connectionism like Dennett (1986, 1991), Smolensky (1988), Pinker & Prince (1988), and Rumelhart & McClelland (1985) follow Holyoak & Thagard (1995) in their belief that, since the workings of the mind “are not directly observable, theorizing about them is necessarily analogical: we form hypotheses about mental operations by comparing them to processes that we can more directly observe.  Computer programs are fully inspectable by us, so we can know the data structures they use to represent information, as well as the algorithms that process information by operating on those structures… Cognitive theories inspired by the computational analogy hypothesize first a set of representational structures and second a set of computational processes that operate on those structures to produce intelligent behavior” (p. 238-9).  This argument trades on the analogy between computational processes and mental processes.  For this connectionist analogical argument to be made stronger, it would be necessary to show that the brain and computational hardware are very similar to one another.  Interestingly enough, when we do in fact investigate the relationship between the neuronal wiring of the brain and the nodal wiring of computer hardware, we notice some fundamental differences between the two.

The first difference between the brain and computational hardware has to do with the fact that there are several types of neurons having a variety of functions in the brain, whereas this is not the case for the computational nodes at work in, for example, Thagard, Holyoak, Nelson, & Gochfeld’s (1990) ARCS (Analog Retrieval by Constraint Satisfaction) and Holyoak & Thagard’s (1989) ACME (Analogical Mapping by Constraint Satisfaction) models.  Stellar, pyramidal, spindle, and chandelier neurons have a variety of responses, as well as differing specialized functions in the brain (Kandel, Schwartz, & Jessell, 2000; Churchland, 1986).  By contrast, computational nodes in ARCS and ACME follow one basic type of functioning having to do with what is known as the on-off logic gate.

The on-off logic gate refers to the fact that computational nodes only can respond to information in one of two ways, having to do with either an on-off, yes-no, or 1-0 response, depending upon the program.  Early on in the history of cognitive science, the computer’s logic gates and the brain’s neurons both were thought to follow this basic pattern; the neuron was conceived as either on (firing) or off (not firing) (see Turing, 1950; von Neumann, 1958; Jackendoff, 1987; Holyoak & Thagard, 1995; Copeland, 1993).  However, now we know that whereas nodal logic gates are either on or off, neurons are never fully off, but are said to maintain a resting potential prior to their inhibitory or excitatory activity.  In a sense, neurons are always on to a certain degree (Kandel et al., 2000).  This represents a second way in which the brain and computational hardware are different from one another.

There are other ways in which neurons and nodes are different from one another.  (A) Neurons (and cells in general) have the properties of internal-hierarchical self-maintenance and homeostasis, whereas nodes do not.  The organelles of a cell are specialized in their activities and organized in such a way so as to preserve the stability of the cell (Smolensky, 1988; Rumelhart & McClelland, 1985).  (B) Kandel et al. (2000) and Pinker & Prince (1988) have demonstrated that the speed at which neurons process information is much slower than that of nodal connections.  (C) A neuron can have several thousand dendritic connections with other neurons, whereas a typical node can have only several connections at most (Kandel et al., 2000; Smolensky, 1988; Rumelhart & McClelland, 1985).  (D) Unlike computational networks that have specific locations for memory, in mammals memory is not associated with any specific area of the brain, but is distributed throughout the brain (Kandel et al., 2000; Pinker & Prince, 1988).  (E) Finally, the brain is highly adaptable in the sense that certain groups of neurons can perform alternate functions in the event of brain damage.  There is a kind of flexibility or malleability present in neuronal networks that is not found in nodal networks (Kandel et al., 2000; Copeland, 1993; Searle, 1992; Fodor, 2001; Churchland, 1986).

Now, simply to point out the differences between neurons and nodes is not enough to show the deficiencies in the computational approach.  We need to show that these differences actually make a difference in terms of an organism’s ability to creative problem solve, such that if the organism did not have a biological structure with a specific evolutionary history, then it could not effectively creatively problem solve.  In other words, we need to show that the biological particularities, complete with their evolutionary history, matter to creative problem solving.

First of all, a biological organ like the brain, its genetic make-up, and its environment are involved in a causal interplay, aiding each other in evolving and developing.  This dynamic relationship of organism, genes, and environment is what thinkers refer to as epigenesis (e.g., Berra, 1990).  Conscious brains affect environments and biology, and environments and biology affect conscious brains.  This microcosmic interrelation is representative of interrelations present at a macrocosmic level in all of nature.  All of nature is pliable, in this epigenetic sense, and this is a fundamental insight that Darwin made in rejecting the stasis of organisms.

Somewhat paradoxically (or ironically), if it were not for the fortuitous genetic mutations within these specific biological entities and the specific environmental shifts that make up the specific evolutionary history of these specific biological entities, organisms would never have evolved light/dark sensitivity areas, brains, mental states, and then abilities to creatively problem solve.  At the same time, if organisms did not have an epigenetic history whereby they engaged in some forms of proto-typical brain process functioning and problem solving, they never would have been able to eventually creatively negotiate environments so as to survive and flourish.  This interplay is one whereby genes affect brains, environment affects brains, brain affects genes, brains affects environments, etc., in this huge epigenetic cycle.

The neural circuitry necessary for creative problem solving in humans evolved for the purposes of negotiating specific organic environments.  Traits (e.g., neurons, organs, psychological states) develop in evolutionary history to function as a result of chance mutations and the natural selection of the trait that is most fit, given the particular environment in which the trait exists.  The varieties of neurons in an animal, complete with their specialized functioning, testify to the influence of the environment.  There are olfactory receptor neurons specialized to process odors, auditory receptor neurons specialized to process sound waves (both through processes of transduction), motor neurons specialized to produce muscular contractions, interneurons facilitating speed of signal between and among neurons, etc.  In the visual system, some neurons are responsive to colors, some are responsive to movement, while other are responsive to lines in specific orientations, faces, stereoptic, and several other features of objects in the environment.  Neurons come in a wide variety of types, having a variety of specialized functions and differing greatly in terms of their size, axonal length, and characteristic pattern of dendritic arborization.  Their myriad forms and functions are dependent upon internal and external environmental factors (Kandel et al., 2000; Churchland, 1986).

The Gestalt psychologists noted that the visual system has built-in mechanisms whereby the visual scene is grouped according to the following principles: closure, the tendency of the visual system to ignore small breaks or gaps in objects; good continuation, the tendency to group straight or smoothly curving lines together; similarity, the tendency to group objects of similar texture and shape together; and proximity, the tendency to group objects that are near to one another together.  Numerous studies have ratified these principles as reflective of the visual system (Wertheimer, 1912, 1923; Kanizsa, 1976, 1979; Peterhans & von der Heydt, 1991; Gray, 1999; Sekuler & Blake, 2002).  These mechanisms can work only if the environment actually displays the features on which the visual system is capitalizing.  There must be these kinds of regularities out there in the world, or else it seems these principles would not be able to be delineated (Brunswik & Kamiya, 1953).  Over 100 years ago, James (1892) noted that mind and world “have evolved together, and in consequence are something of a mutual fit” (p. 4), and the Gestalt principles underscore this mutual fit.

Events in the physical environment are composed of materials that the human brain, complete with its specialized modules, has evolved to perceive and discriminate.  These materials take forms ranging from specific chemicals, to mechanical energy, to electromagnetic radiation (light), and are discriminated by the different sensory modalities that are specifically attuned to these stimuli.  The important point to note is that the various specialized modules never would have come to be if it were not for the specific organic environment in which the organism found itself.

Likewise, the mind contains specific mental modules that evolved in our past to solve specific problems of survival, such as face recognition, mental maps, intuitive mechanics, intuitive biology, kinship, language acquisition, mate selection, and cheating detection – again – given the particular environment in which the early hominin existed.  Adaptive problems are “problems that are specifiable in terms of evolutionary selection pressures, i.e., recurring environmental conditions that affect, or have affected, the reproductive success of individual organisms” (Wheeler & Atkinson, 2001, p. 242).  It was the specific environmental pressures of the Pleistocene that acted as the conditions for the possibility of creative problem solving.  The successful progression from the typical jungle environments to the atypical and novel savannah-type environments of our early hominin ancestors was the occasion for a mental capacity to emerge that creatively could handle the new environment.  Those specific organic/natural conditions make all the difference when describing conscious creative problem solving in humans.

It seems fundamentally misguided to investigate mental processes wholly divorced from the brain on which these mental processes are realized.  Churchland (1986) makes at least four points that dry-mind computational functionalists should keep in mind:

1. Our mental states and processes are states and processes of our brains.  2. The human nervous system evolved from simpler nervous systems.  3. Brains are by far the classiest information processors available for study.  In matters of adaptability, plasticity, appropriateness of response, motor control, and so forth, no program has ever been devised that comes close to doing what brains do – not even to what lowly rats do.  If we can figure out how brains do it, we might figure out how to get a computer to mimic how brains do it.  4. If neuroscientists are working on problems of memory and learning, studying cellular changes, synaptic changes, the effects of circumscribed lesions, and so forth, it is perverse for a cognitive scientist trying to understand memory and learning to ignore systematically what the neuroscientists have discovered. (p. 362)

Further, Sober (1990) titles one of his articles “Putting the function back into functionalism,” noting that the mind’s information processing (functionalism) should not be wholly divorced from the biological processes (functions) upon which this information depends (cf. Ruse, 1971, 1973).  It seems that all three of Marr’s levels of analysis of information processing should be taken into consideration when investigating a biologically-based entity’s mental functioning.  The specific hardware of the brain does make a difference to the processing of information when we consider the specific evolutionary history of the brain.

Connectionism and problem solving

The issue of the hardware of the brain and computational networks aside, as Churchland intimates in the quotation above, another important test of intelligence has to do with the behaviors that organisms and computational mechanisms exhibit.  Ever since Turing’s (1950) test, in conjunction with experiments performed in behavioral psychology, one of the ways in which we can judge whether some thing is intelligent is by how well it solves problems.  We run worms and rats through mazes to solve problems in order to get a sense of their intelligence level.  We also run robots through police and military tests to see if they can solve the problems of not being shot or blown up by a bomb.  The problem for connectionists (and other functionalists alike) is that such robots, running programs based on computational models, consistently fail at even the simplest of routine problem solving exercises  (Gerkey & Matari, 2002; Franz, 2003; Moravec, 1999a, 1999b, 2000; cf. Brooks, 1991).  Minimally, this failure calls the legitimacy of the computational approach into question.  After discussing the fact that we “still don’t have the fabled machine that can make breakfast without burning down the house” or “even one that can learn anything much except statistical generalizations,” Fodor (2001) notes that the “failure of our AI is, in effect, the failure of the Classical Computational Theory of the Mind to perform well in practice.  Failures of a theory to perform well in practice are much like failures to predict the right experimental outcomes (arguably, indeed, the latter is a special case of the former)” (pp. 37-8).

Moravec (1999a, 1999b, 2000) speculates that by 2010, robots with connectionist networks of 5,000 MIPS (million-instructions-per-second) will achieve the cognitive status comparable to a lizard.  Provided the money is there for further research, such robots will be followed by mouselike (100,000 MIPS), monkeylike (5 million MIPS), and humanlike (100 million MIPS) robots in generations to come.  Now, is it possible that one day in the future artificial beings will become creative problem solvers?  Yes.  However, I believe along with Searle (1992) that this would be a reality only if we could simulate biological processes in other material media.  Is it also possible that other beings on other planets lacking neuro-matter could be conscious?  Yes.  However, what matters is how it is that human beings on this planet in this world have become conscious creative problem solvers.  This is why I think it is invaluable to investigate the biology and evolution of the mind.

Why is it that robots having dry-mind computational networks break down while trying to complete minimally complex tasks in routine situations (Finke et al., 1992; Churchland, 1993; Dreyfus, 1992; Moravec, 1999a, 1999b, 2000)?  Part of the answer is that they lack the conscious abilities to flexibly select and integrate information.  They just do not fall into the category of things having an evolutionary history, out of which these conscious abilities emerged.  If such computationally-based mechanisms cannot engage successfully in routine forms of problem solving, then how could they engage successfully in nonroutine creative forms of problem solving?  Do we really think that, one day, computationally-based mechanisms will be able to make the kind of bissociative connection that Dr. Wells made concerning the use of nitrous oxide?

A lack of consciousness – understood as Mithen’s form of cognitive fluidity – is a good candidate for explaining why, for example, ants can get thrown off track so easily by introducing novel stimuli in their environments (Moravec, 1999b; McFarland & Bosser, 1993), moths get confused and kill themselves in flames or luminescent bug-killers (Moravec, 1999b; McFarland & Bosser, 1993), connectionist networks are unable to distinguish syntax from semantics (Searle, 1990), the robotic eye cannot discern that the following are all the same word: CAT, Cat, Cat, Cat, Cat (Copeland, 1993), and robots continually blow themselves up in military training exercises while trying to negotiate environments as they are being bombarded by novel stimuli (Moravec, 1999a, 2000; also Churchland, 1993; Fodor, 2001).

Following Sternberg (2001), it seems that processes for deciding what to do, learning what to do, and learning how to do – viz., those processes rightly associated with unconscious modularity (as is present in ants and moths) and/or computational efficiency (as is present in a pre-programmed robot) – are not enough in negotiating this incredibly complex and diverse world.  A conscious process (again, like the one Mithen envisions) seems necessary in our problem solving, whether we attempt to cope with new situations, select new environments when old ones become unsatisfactory, creatively re-use information from old environments in new ones, coherently set up goals via the selectivity and integration of information, synthesize disparate concepts, construct theories to explain phenomena, or invent tools based upon mental blueprints (Crick, 1994; Searle, 1992; Finke et al., 1992).

I believe that if an artificially intelligent being with a dry mind is to become creative, such a being must be conscious.  But consciousness is a functionally-emergent property of the brain of the human species specifically, and brains are organs with an epigenetic history (which includes the interplay of genotypic and phenotypic characteristics abiding by evolutionary principles).  My argument goes, then, that if an artificially intelligent being is going to become creative, it will need some kind of epigenetic history.  By transposition: if a being does not have an epigenetic history, then it cannot be a biological organism having a brain; if such a being lacks a brain, then it cannot be conscious; and without consciousness, a being cannot creatively problem solve.  I think this lack of an evolutionary history is what accounts for many of the problems thinkers have with the dry-mind computational approach to understanding the mind of a biological entity (see Copeland, 1993; Dreyfus, 1992; Kosslyn & Koenig, 1995; Zornetzer, Davis, & Lau, 1990; Churchland & Sejnowski, 1992; McFarland & Bosser, 1993; Born, 1987).

The philosophical upshot of my investigation of the relationship between dry-mind and wet-mind is to make clear that wet-mind biological and evolutionary approaches have amassed enough information such that dry-mind approaches can now take cues from these burgeoning wet-mind sciences.  It used to be that functionalists and other cognitive scientists were looked to as having an accurate picture of how the mind works, while biologists and evolutionists were looked to secondarily as having no real philosophical say in mental phenomena (Putnam, 1960; Fodor, 1983; Barlow, 1994; Lycan, 1995; Block, 1994).  Interestingly enough, this kind of functionalist view undermines its own intentions to be grounded materialistically, since the mind is viewed as a new kind of computational or linguistic epiphenomenon that does not look to matter, viz., the brain!!!, for its identity.

However, with the advances made in the neurobiological and evolutionary sciences, dry-mind can (and really should) look to wet-mind for some guidance.  Following Chuchland (1986), “if we can figure out how brains do it, we might figure out how to get a computer to mimic how brains do it” (p. 362).  In fact, the move cognitive scientists have made into parallel distributed processing (PDP) demonstrates that cognitive science is looking to neuroscience in trying to explain the workings of the mind – rather than the other way around – since PDPs are set up to reflect actual neural networks more accurately.  As was noted above, connectionists model their networks after the tripartite neuronal system of the brain.

This is not to say that work in cognitive science and artificial intelligence is unimportant; to the contrary, we have learned much about the brain, mind, human behavior, and evolution because of such research (e.g., Bechtel & Abrahamsen, 2002; Lek & Guegan, 2000; Hinton & Nowlan, 1987; Barto, 1985; Smith, 1987; Christiansen & Chater, 1994).  This is just to say that cognitive scientists need to take information, cues, and clues from the biological, psychological, social, and evolutionary sciences.  In fact, for years researchers in cognitive science and artificial intelligence have been utilizing evolutionary principles when they construct their own simulated mental apparatuses and environments.

For example, Holland (1975) constructed a simple computational program on the model of evolutionary principles that would learn by “evolving” and “adapting” to new information the program received.  He was concerned with utilizing evolutionary ideas in computational systems with the ultimate goal of understanding natural systems more completely.  His insight was that an organized system must come together part-by-part with the separate parts, at first, functioning so as to solve a certain distinct problem.  Eventually, more complex levels could be added to the more basic levels of his computational program, in much the same way more complex levels are added to, or evolve in, organic hierarchies.  This became the model for how computer networks are built up from the fundamental ifs and thens or the 1s and 0s, to the more complexly functioning Big Blues or world-wide-webs (as in Sperber, 1994; Moravec, 1999a; Johnson-Laird, 1988).  However, the important point to realize is that this is also the way natural/historical processes appear to work in evolutionary advances (see Deacon, 1997; Berra, 1990; Cosmides & Tooby, 1992, 1994; Mithen, 1996).

In an article entitled “Evolving parallel computation,” Thearling & Ray (1997) have applied a natural selection model to their software system they call TIERRA.  TIERRA simulated the replication of multi-cellular organisms in complex and changing environments over several generations.  At first, the virtual organisms processed information in a slower, if-then, serial fashion.  Their findings were impressive because, through several generations and in several changing environments, the virtual multi-cellular organisms were able to “evolve” parallel processing mechanisms to deal with the information they were receiving from their virtual environments.  Other researchers have utilized evolutionary principles as a starting point for their computational models, whether it be to understand how computers can learn (Williams, 1988; Nolfi 1990; Yao & Liu, 1997, 1998; French & Sougne, 2001), exchange data (Lerman & Rudolph, 1994), discriminate in-coming information in some virtual environment (Menczer & Belew, 1994), devise a primitive language (Hadley, 1994), solve problems (Koza, 1990) or even play checkers (Chellapilla & Fogel, 2001).

So it is understood by several researchers that, just as organic processes do what they do because they have been selected for in an evolutionary history (e.g., Goodale & Murphy, 2000; Kosslyn & Koenig, 1995; Allman, 2000; Desimone et al., 1984; Casagrande & Kaas, 1994; Edelman & Tononi, 2000; Shallice, 1997; Marr, 1983; Sereno et al., 1995), so too, computational systems can be modeled on similar virtual or simulated “evolutionary” histories.  And further, the information gleaned from these simulated tests can help in understanding the natural world, of which the mind is a part.

The end result is that biologists and evolutionists can now be looked to as contributing significant pieces to the puzzle concerning the workings of the mind.  When all is said and done, I support Churchland’s (1993) claim that cognitive science should not be autonomous with respect to neuroscience, psychology, and the other empirical sciences.  I endorse Fodor’s (1998) observation that archeology and the biological sciences are good places to uncover the nature of the mind.  I concur with Pinker (1994), echoing Chomsky, that if research in artificial intelligence is to study effectively the mind, then it needs “constant integration across disciplinary lines” (p. 15).  Finally, I agree with Donald (1997) that the “problem of cognitive evolution demands the widest possible range of information, (from) neurolinguistics, anthropology, paleontology, neuroanatomy, and especially cognitive psychology” (p. 356).

Acknowledgements

I wish to thank George Terzis, Brian Cameron, Eric LaRock, and Benoit Hardy-Vallee for comments on earlier versions of this paper.

References

Abravanel, E. (1991). Does immediate imitation influence long-term memory for observed actions? Journal of Experimental Child Psychology, 21, 614-23.

Allman, J. (2000). Evolving brains. New York: Scientific American Library.

Arp, R. (2005a). Scenario visualization: One explanation of creative problem solving. Journal of Consciousness Studies, 12, 31-60.

_____, (2005b). Selectivity, integration, and the psycho-neuro-biological continuum. Journal of Mind and Behavior, 6&7, 35-64.

_____, (2006a). Awareness and consciousness: Switched-on rheostats. Journal of Consciousness Studies, forthcoming.

_____, (2006b). The environments of our Hominin ancestors, tool usage, and scenario visualization. Biology & Philosophy, 21, 95-117.

_____, (2007a). Scenario visualization: An evolutionary account of vision-related creative problem solving. Cambridge, MA: MIT Press.

_____, (2007b). An integrated approach to the philosophy of mind. Cambridge: Cambridge University Press, in production.

Barlow, H. (1994). What is the computational goal of the neocortex? In C. Koch & J. Davis (Eds.), Large-scale neuronal theories of the brain (pp. 1-22). Cambridge, MA: MIT Press.

Barto, A. (1985). Learning by statistical co-operation of self-interested neuron-like computing elements. Human Neurobiology, 4, 229-56.

Bechtel, W., & Abrahamsen, A. (1991). Connectionism and the mind: An introduction to parallel processing in networks. New York: Blackwell.

_____, (2002). Connectionism & the mind: Parallel processing dynamics and evolution. New York: Blackwell.

Berra, T. (1990). Evolution and the myth of creationism. Stanford: Stanford University Press.

Bitterman, M. (1965). The evolution of intelligence. Scientific American, 212, 92-100.

_____, (1975). The comparative analysis of learning. Science, 188, 699-709.

Block, N. (1980). Introduction: What is functionalism? In N. Block (Ed.), Readings in philosophy and psychology (pp. 171-84). Cambridge, MA: Harvard University Press.

_____, (1994). Functionalism. In S. Guttenplan (Ed.), A companion to the philosophy of mind (pp. 315-32). New York: Blackwell Publishers.

Boden, M. (1990). The creative mind: Myths and mechanisms. New York: Basic Books.

Bogdan, R. (1994). Grounds for cognition: How goal-guided behavior shapes the mind. Hillsdale, CA: LEA Publishers.

Born, R. (Ed.), (1987). Artificial intelligence: The case against. London: Croom Helm.

Brooks, R. (1991). Intelligence without reason. Proceedings of the Twelfth International Joint Congress on Artificial Intelligence, San Mateo, California, 12, 569-95.

Brunswik, E., & Kamiya, J. (1953). Ecological cue-validity of proximity and other Gestalt factors. American Journal of Psychology, 66, 20-32.

Buller, D. (2005). Adapting minds: Evolutionary psychology and the persistent quest for human nature. Cambridge, MA: MIT Press.

Buss, D. (1999). Evolutionary psychology: The new science of the mind. Boston: Allyn and Bacon.

Byrne, R. (1995). The thinking ape: Evolutionary origins of intelligence. Oxford: Oxford University Press.

Call, J., & Tomasello, M. (1994). The social learning of tool use by orangutans (Pan pygmaeus). Human Evolution, 9, 297-313.

Calvin, W. (2004). A brief history of the mind: From apes to intellect and beyond. Oxford: Oxford University Press.

Casagrande, V., & Kaas, J. (1994). The afferent, intrinsic, and efferent connections of primary visual cortex in primates. In A. Peters & K. Rockland (Eds.), Cerebral cortex, volume 10: Primary visual cortex in primates (pp. 201-59). New York: Plenum Press.

Chellapilla, K., & Fogel, D. (2001). Evolving an expert checkers playing program without using human expertise. IEEE Transactions on Evolutionary Computation, 5, 422-28.

Christiansen, M., & Chater, N. (1994). Generalisation and connectionist language learning. Mind and Language, 9, 273-87.

Churchland, P. S. (1986). Neurophilosophy: Toward a unified science of the mind-brain. Cambridge, MA: MIT Press.

_____, (1993). The co-evolutionary research ideology. In A. Goldman (Ed.), Readings in philosophy and cognitive science (pp. 745-67). Cambridge, MA: MIT Press.

_____, (1997). Can neurobiology teach us anything about consciousness? In N. Block et al. (Eds.), The nature of consciousness (pp. 127-40). Cambridge, MA: MIT Press.

Churchland, P. S., & Sejnowski, T. (1992). The computational brain: Models and methods on the frontiers of computational neuroscience. Cambridge, MA: MIT Press.

Clayton, N., & Krebs, J. (1994). Hippocampal growth and attrition in birds affected by experience. Proceedings of the National Academy of Science, 91, 7410-14.

Copeland, J. (1993). Artificial intelligence. Cambridge, MA: Blackwell.

Cosmides, L., & Tooby, J. (1987). From evolution to behavior: Evolutionary psychology as the missing link. In J. Dupre (Ed.), The latest on the best: Essays on evolution and optimality (pp. 27-36). Cambridge, MA: Cambridge University Press.

_____, (1992). The psychological foundations of culture. In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind (pp. 19-136). New York: Oxford University Press.

_____, (1994). Origins of domain specificity: The evolution of functional organization. In L. Hirschfeld & S. Gelman (Eds.), Mapping the mind: Domain specificity in cognition and culture (pp. 71-97). Cambridge, MA: Cambridge University Press.

Crick, F. (1994). The astonishing hypothesis. New York: Simon & Schuster.

Crick, F., & Koch, C. (1999). The problem of consciousness. In A. Damasio (Ed.), The Scientific American book of the brain (pp. 311-24). New York: Lyons Press.

Davidson, J. (1995). The suddenness of insight. In R. Sternberg & J. Davidson (Eds.), The nature of insight (pp. 7-27). Cambridge, MA: MIT Press.

Deacon, T. (1997). The symbolic species: The co-evolution of language and the brain. New York: W.W. Norton and Company.

Dennett, D. (1986). Content and consciousness. London: Routledge & Kegan Paul.

_____, (1991). Consciousness explained. Boston: Little, Brown and Company.

Desimone, R., Albright, T., Gross, C., & Bruce, C. (1984). Stimulus-selective properties of inferior temporal neurons in the macaque. Journal of Neuroscience, 4, 2051-62.

Diamond, M. (1988). Enriching heredity. New York: The Free Press.

Diamond, M., & Hopson, J. (1989). Magic trees of the mind. New York: Dutton Press.

Dickinson, A. (1980). Comparative animal learning theory. Cambridge, UK: Cambridge University Press.

Dominowski, R. (1995). Productive problem solving. In S. Smith, T. Ward, & R. Finke (Eds.), The creative cognition approach (pp. 73-96). Cambridge, MA: MIT Press.

Donald, M. (1991). Origins of the modern mind. Cambridge, MA: Harvard University Press.

_____, (1997). The mind considered from a historical perspective. In D. Johnson & C. Erneling (Eds.), The future of the cognitive revolution (pp. 355-65). New York: Oxford University Press.

Dreyfus, H. (1992). What computers still can’t do: A critique of artificial reason. Cambridge: MIT Press.

Edelman, G., & Tononi, G. (2000). Reentry and the dynamic core: Neural correlates of conscious experience. In T. Metzinger (Ed.), Neural correlates of consciousness (pp. 139-56). Cambridge, MA: MIT Press.

Emery, N., & Clayton, N. (2004). The mentality of crows: Convergent evolution of intelligence in corvids and apes. Science, 306, 1903-1907.

Feldman, J., & Ballard, D. (1982). Connectionist models and their properties. Cognitive Science, 6, 205-54.

Fiorito, G., et al. (1990). Problem solving ability of Octopus vulgaris Lamark (mollusca, cephalopodo). Behavior and Neural Biology, 53, 217-30.

Finke, R., Ward, T., & Smith, S. (1992). Creative cognition: Theory, research and applications. Cambridge, MA: MIT Press.

Fodor, J. (1983). The modularity of mind. Cambridge, MA: MIT Press.

_____, (1985). Précis of ‘The Modularity of Mind’. The Behavioral and Brain Sciences, 8, 1-42.

_____, (1998). In critical condition: Polemical essays on cognitive science and the philosophy of mind. Cambridge, MA: MIT Press.

_____, (2001). The mind doesn’t work that way: The scope and limits of computational psychology. Cambridge, MA: MIT Press.

Frank, E., & Wenner, P. (1993). Environmental specification of neuronal connectivity. Neuron, 10, 779-85.

Franz, M. (2003). Robots with cognition? Tübinger Wahrneh mungs konferenz, 6, Proc. 38.

French, R., & Sougne, J. (2001). Connectionist models of learning, development & evolution. New York: Springer.

Gardner, H. (1993). Multiple intelligences: The theory in practice. New York: Basic Books.

Gerkey, B., & Matari, M. (2002). Pusher-watcher: An approach to fault-tolerant tightly-coupled robot coordination. Proceedings of the IEEE International Conference on Robotics and Automation ICRA, 2002, 464-9.

Goodale, M., & Murphy, K. (2000). Space in the brain: Different neural substrates for allocentric and egocentric frames of reference. In T. Metzinger (Ed.), Neural correlates of consciousness (pp. 189-202). Cambridge, MA: MIT Press.

Gray, C. (1999). The temporal correlation hypothesis of visual feature integration: Still alive and well. Neuron, 24, 31-47.

Hadley, R. (1994). Systematicity in connectionist language learning. Mind and Language, 9, 247-71.

Hale, M. (1989). Mechanisms of the mind: An evolutionary perspective. New York: Hale-Van Ruth.

Hall, G. (1994). Pavlovian conditioning: Laws of association. In N. Mackintosh (Ed.), Animal learning and cognition (pp. 15-43). San Diego: Academic Press.

_____, (1996). Learning about associatively activated stimulus representations: Implications for acquired equivalence and perceptual learning. Animal Learning and Behavior, 24, 233-55.

Heyes, C., & Dawson, G. (1990). A demonstration of observational learning in rats using a bidirectional control. Quarterly Journal of Experimental Psychology, 42B, 59-71.

Heyes, C., Jaldon, E., & Dawson, G. (1992). Imitation in rats: Initial responding and transfer evidence from a bidirectional control procedure. Quarterly Journal of Experimental Psychology, 45B, 229-40.

Hinton, G., & Nowlan, S. (1987). How learning can guide evolution. Complex Systems, 1, 495-502.

Hirschfeld, L., & Gelman, S. (1994). Toward a topography of mind: An introduction to domain specificity. In L. Hirschfeld & S. Gelman (Eds.), Mapping the mind: Domain specificity in cognition and culture (pp. 3-35). Cambridge, UK: Cambridge University Press.

Holland, J. (1975). Adaptation in natural and artificial systems. Ann Arbor: University of Michigan Press.

Holloway, M. (2003). The mutable brain. Scientific American, 289, 78-85.

Holyoak, K., & Thagard, P. (1989). Analogical mapping by constraint satisfaction. Cognitive Science, 13, 295-355.

_____, (1995). Mental leaps: Analogy in creative thought. Cambridge, MA: MIT Press.

Humphrey, N. (1992). A history of the mind: Evolution of the birth of consciousness. New York: Simon & Schuster.

Hunter, W. (1913). The delayed reaction in animals and children. Behavior Monographs, 2, 1-86.

Jackendoff, R. (1987). Consciousness and the computational mind. Cambridge, MA: MIT Press.

James, W. (1890). The principles of psychology. New York: Henry Holt and Company.

Johnson-Laird, P. (1988). The computer and the mind: An introduction to cognitive science. Cambridge, MA: Harvard University Press.

Kandel, E., Schwartz, J. & Jessell, T. (Eds.), (2000). Principles of neural science. New York: McGraw-Hill.

Kanizsa, G. (1976). Subjective contours. Scientific American, 234, 48.

_____, (1979). Organization in vision: Essays on gestalt perception. New York: Praeger.

Kimura, D. (1989). Left hemisphere control of oral and brachial movements and their relation to communication. Philosophical Transactions of the Royal Society, Series B, B292, 135-49.

Koestler, A. (1964). The act of creation. New York: Dell.

Kosslyn, S., & Koenig, O. (1995). Wet mind: The new cognitive neuroscience. New York: The Free Press.

Koza, J. (1990). Genetic Programming: A paradigm for genetically breeding populations of computer programs to solve problems. Technical Report STAN-CS-90-1314, Stanford University.

Lek, S., & Guegan, J. (2000). Artificial neuronal networks: Application to ecology & evolution. New York: Telos Press.

Lerman, G., & Rudolph, L. (1994). Parallel evolution of parallel processors. The Hague: Kluwer Academic Publishers.

Lycan, W. (1995). Consciousness. Cambridge, MA: MIT Press.

Mackintosh, N. (1983). Conditioning and associative learning. Oxford: Clarendon Press.

_____, (1995). Categorization by people and by pigeons. Quarterly Journal of Experimental Psychology, 48B, 193-214.

Macphail, E. (1996). Cognitive function in mammals: The evolutionary perspective. Cognitive Brain Research, 3, 279-90.

_____, (1998). The evolution of consciousness. New York: Oxford University Press.

Mahner, M., & Bunge, M. (2001). Function and functionalism: A synthetic perspective. Philosophy of Science, 68, 75-94.

Maier, N. (1932). A study of orientation in the rat. Journal of Comparative Psychology, 14, 387-99.

Mandler, G. (1995). Origins and consequences of novelty. In S. Smith et al. (Eds.), The creative cognition approach (pp. 9-26). Cambridge, MA: MIT Press.

Marr, D. (1983). Vision. New York: Freeman Press.

Mattson, B., Sorensen, J., Zimmer, J., & Johansson, B. (1997). Neural grafting to experimental neocortical infarcts improves behavioral outcome and reduces thalamic atrophy in rats housed in enriched but not in standard environments. Stroke, 28, 1225-31.

Mayer, R. (1995). The search for insight: Grappling with Gestalt psychology’s unanswered questions. In R. Sternberg & J. Davidson (Eds.), The nature of insight (pp. 3-32). Cambridge, MA: MIT Press.

McFarland, D., & Bosser, T. (1993). Intelligent behavior in animals and robots. Cambridge, MA: MIT Press.

Menczer, F., & Belew, R. (1994). Evolving sensors in environments of controlled complexity. In R. Brooks (Ed.), Proceedings of the Fourth Conference on Artificial Life (pp. 33-45). Cambridge, UK: Polity Press

Miles, H., Mitchell, R., & Harper, S. (1996). Simon says: The development of imitation in an enculturated orangutan. Journal of Comparative Psychology, 105, 145-60.

Mitchell, R. (1987). A comparative developmental approach to understanding imitation. Perspectives in Ethology, 7, 183-215.

_____, (1993). Mental models of mirror self-recognition: Two theories. New Ideas in Psychology, 11, 295-325.

Mithen, S. (1996). The prehistory of the mind: The cognitive origins of art, religion and science. London: Thames and Hudson.

_____, (1999). Handaxes and ice age carvings: Hard evidence for the evolution of consciousness. In S. Hameroff, A. Kaszniak, & D. Chalmers (Eds.), Toward a science of consciousness: The third Tucson discussions and debates (pp. 281-96). Cambridge, MA: MIT Press.

Moore, B. (1992). Avian movement imitation and a new form of mimicry: Tracing the evolution of a complex form of learning. Behavior, 122, 231-63.

Moravec, H. (1999a). Robot: Mere machine to transcendent mind. New York: Oxford University Press.

_____, (1999b). Rise of the robots. Scientific American, 122, 124-35.

_____, (2000). Robots, re-evolving mind (pp. 1-9). On-line article: www.frc.ri.cmu.edu/~hpm/project.archive/robot.papers/2000/Cerebrum

Nolfi, S. (1990). Learning and evolution in neural networks. CRL Technical Report 9019, University of California, San Diego.

Olton, D. & Samuelson, R. (1976). Remembrance of places passed: Spatial memory in rats. Journal of Experimental Psychology: Animal Behavior Processes, 2, 97-116.

Palmer, J., & Palmer, A. (2002). Evolutionary psychology: The ultimate origins of human behavior. Needham Heights, MA: Allyn and Bacon.

Parker, S. (1996). Apprenticeship in tool-mediated extractive foraging: The origins of imitation, teaching and self-awareness in great apes. Journal of Comparative Psychology, 106, 18-34.

Patterson, K., & Wilson, B. (1987). A ROSE is a ROSE or a NOSE: A deficit in initial letter identification. Cognitive Neuropsychology, 7, 447-77.

Pearce, J. (1997). Animal learning and cognition. Erlbaum, UK: Psychology Press.

Peterhans, E., & von der Heydt, R. (1991). Subjective contours – bridging the gap between psychophysics and physiology. Trends in Neuroscience, 14, 12.

Pinker, S. (1994). The language instinct. New York: W. Morrow and Company.

_____, (1997). How the mind works. New York: W. W. Norton & Company.

_____, (2002). The blank slate: The modern denial of human nature. New York: Penguin Books.

Pinker, S., & Prince, A. (1988). On language and connectionism: Analysis of a parallel distributed processing model of language acquisition. In S. Pinker & J. Mehler (Eds.), Connections and symbols (pp. 73-193). Cambridge, MA: MIT Press.

Plotkin, H. (1997). Evolution in mind: An introduction to evolutionary psychology. Cambridge, MA: Harvard University Press.

Putnam, H. (1960). Minds and machines. In S. Hook (Ed.), Dimensions of mind (pp. 27-37). New York: Collier Books.

Pylyshyn, Z. (1980). Computation and cognition: Issues in the foundation of cognitive science. Behavioral and Brain Sciences, 3/1, 111-34.

Receveur, H., & Vossen, J. (1998). Changing rearing environments and problem solving flexibility in rats. Behavioral Processes, 43, 193-210.

Reiner, A., Perkel, D., Mello, C., & Jarvis, E. (2004). Songbirds and the revised avian brain nomenclature. Annals of the New York Academy of Sciences, 1016, 77-108.

Rescorla, R. (1988). Behavioral studies of Pavlovian conditioning. Annual Review of Neuroscience, 11, 329-52.

Roberts, A., Robbins, T., & Weiskrantz, L. (1998). The prefrontal cortex: Executive and cognitive functions. Oxford: Oxford University Press.

Roberts, R. (1989). Serendipity: Accidental discoveries in science. New York: John Wiley & Sons, Inc.

Roberts, W., & Grant, D. (1974). Short-term memory in the pigeon with presentation time precisely controlled. Learning and Motivation, 5, 393-408.

Rumelhart, D., & McClelland, J. (1985). PDP models and general issues in cognitive science. In D. Rumelhart & J. McClelland (Eds.), Parallel distributed processing: Explorations in the microstructure of cognition, volume 1: foundations (pp. 110-46). Cambridge, MA: MIT Press.

Ruse, M. (1971). Functional statements in biology. Philosophy of Science, 38, 87-95.

_____, (1973). The philosophy of biology. London: Hutchinson.

Russon, A., & Galdikas, B. (1993). Imitation in free-ranging rehabilitant orangutans. Journal of Comparative Psychology, 107, 147-61.

_____, (1995). Constraints on great apes’ imitation: Model and action selectivity in rehabilitant orangutan (Pan pygmaeus) imitation. Journal of Comparative Psychology, 109, 5-17.

Savage-Rumbaugh, E., & Boysen, S. (1978). Linguistically-meditated tool use and exchange by chimpanzees (Pan troglodytes). Brain and Behavioral Science, 4, 539-53.

Sekuler, R., & Blake, R. (2002). Perception. New York: McGraw Hill.

Scher, S., & Rauscher, F. (2003). Nature read in truth or flaw: Locating alternatives in evolutionary psychology. In S. Scher & F. Rauscher (Eds.), Evolutionary psychology: Alternative approaches (pp. 1-30). Boston: Kluwer Academic Publishers.

Schwartz, J., & Begley, S. (2002). The mind and brain: Neuroplasticity and the power of mental force. New York: Harper Collins.

Searle, J. (1990). Is the brain a digital computer? Proceedings and Addresses of the American Philosophical Association, 64, 21-37.

_____, (1992). The rediscovery of the mind. Cambridge, MA: MIT Press.

Sereno, M., et al., (1995). Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. Science, 268, 889-93.

Shallice, T. (1988). From neuropsychology to mental structure. Cambridge, UK: Cambridge University Press.

_____, (1997). Modularity and consciousness. In N. Block et al. (Eds.), The Nature of consciousness (pp. 277-97). Cambridge, MA: MIT Press.

Shettleworth, S. (2000). Modularity and the evolution of cognition. In C. Heyes & L. Huber (Eds.), The Evolution of cognition (pp. 43-60). Cambridge, MA: MIT Press.

Simon, H. (1975). The functional equivalence of problem-solving skills. Cognitive Psychology, 7, 268-88.

Smolensky, P. (1988). On the proper treatment of connectionism. Behavioral and Brain Sciences, 11, 1-23.

Sober, E. (1990). Putting the function back into functionalism. In W. Lycan (Ed.), Mind and cognition (pp. 97-106). Cambridge, MA: Blackwell Publishing.

Sperber, D. (1994). The modularity of thought and the epidemiology of representations. In L. Hirschfeld & S. Gelman (Eds.), Mapping the mind: Domain specificity in cognition and culture (pp. 39-67). Cambridge, MA: Cambridge University Press.

Sternberg, R. (1996). Cognitive psychology. Fort Worth, TX: Harcourt Brace College Publishers.

_____, (2000). Practical intelligence in everyday life. Cambridge, UK: Cambridge University Press.

_____, (2001). Complex cognition: The psychology of human thought. New York: Oxford University Press.

Stringer, C., & Andrews, P. (2005). The complete world of human evolution. London: Thames & Hudson.

Terzis, G. (2001). How crosstalk creates vision-related eureka moments. Philosophical Psychology, 14, 393-421.

Thagard, P., Holyoak, K., Nelson, G., & Gochfeld, D. (1990). Analog retrieval by constraint satisfaction. Artificial Intelligence, 46, 259-310.

Thearling, K., & Ray, T. (1997). Evolving parallel computation. Complex Systems, 10, 1-8.

Tomasello, M. (1990). Cultural transmission in the tool use and communicatory signaling of chimpanzees. In S. Parker & K Gibson (Eds.), “Language” and intelligence in monkeys and apes (pp. 274-311). Cambridge: Cambridge University Press.

Tomasello, M. et al. (1987). Observational learning of tool use by young chimpanzees. Human Evolution, 2, 175-85.

Tomasello, M. et al. (1993). Imitative learning of actions on objects by children, chimpanzees, and enculturated chimpanzees. Child Development, 64, 1688-1705.

Turing, A. (1950). Computing machinery and intelligence. Mind, 59, 433-60.

van Praag, H., Kempermann, G., & Gage, F. (2000). Neural consequences of environmental enrichment. Nature Reviews Neuroscience, 1, 191-8.

von Neumann, J. (1958). The computer and the brain. New Haven, CT: Yale University Press.

Weir, A., Chappell, J., & Kacelnik, A. (2002). Shaping of hooks in New Caledonian crows. Science, 297, 981.

Weisberg, R. (1995). Case studies of creative thinking: Reproduction versus restructuring in the real world. In S. Smith et al. (Eds.), The creative cognition approach (pp. 53-72). Cambridge, MA: MIT Press.

Wertheimer, M. (1912/1961). Experimental studies on the seeing of motion. In T. Shipley (Ed.), Classics in psychology (pp. 1032-88). New York: Philosophical Library.

_____, (1923/1958). Principles of perceptual organization. In D. Beardslee & M. Wertheimer (Eds.), Readings in perception (pp. 115-35). Princeton, NJ: Van Nostrand.

Wheeler, M., & Atkinson, A. (2001). Domains, brains and evolution. In D. Walsh (Ed.), Naturalism, evolution and mind (pp. 239-66). Cambridge, UK: Cambridge University Press.

Whiten, A., & Custance, D. (1996). Studies of imitation in chimpanzees and children. In C. Heyes & B. Galef (Eds.), Social learning in animals: The roots of culture (pp. 291-318). San Diego: Academic Press.

Whiten, A. et al. (1996). Imitative learning of artificial fruit processing in children (Homo sapiens) and chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 110, 3-14.

Whiten, A. et al. (1999). Cultures in chimpanzees. Nature, 399, 682-5.

Wiener, N. (1948). Cybernetics, or control and communication in the animal and the machine. New York: John Wiley & Sons.

Williams, R. (1988). Toward a theory of reinforcement-learning connectionist systems. Technical Reprt NU-CCS-88-3, Northeastern University.

Wright, A. (1989). Memory processing by pigeons, monkeys, and people. Psychology of Learning and Motivation, 24, 25-70.

_____, (1997). Memory of auditory lists by rhesus monkeys (Macaca mulatta). Journal of Experimental Psychology: Animal Behavior Processes, 23, 441-9.

Yao, X., & Liu, Y. (1997). A new evolutionary system for evolving artificial neural networks. IEEE Transactions on Neural Networks 8, 694-713.

_____, (1998). Towards designing artificial networks by evolution. Applied Mathematics and Computation, 91, 83-90.

Yando, R., Seitz, V., & Ziqler, E. (1978). Imitation: A developmental perspective. Hillsdale, NJ: Lawrence Erlbaum.

Zental, T. et al. (1990). Memory strategies in pigeons’ performance of a radial-arm-maze analog task. Journal of Experimental Psychology: Animal Behavior Processes, 16, 358-71.

Zornetzer, S., Davis, J., & Lau, C. (Eds.). (1990). An introduction to neural and electronic networks. New York: Academic Press.