Natural information processing systems

Evolutionary Psychology
human-nature.com/ep – 2006. 4: 434-458
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Original Article

Natural information processing systems

John Sweller, School of Education, University of New South Wales, Sydney NSW 2052, Australia
Email: j.sweller@unsw.edu.au

Susan Sweller, School of Education, University of New South Wales, Sydney NSW 2052, Australia

Abstract: Natural information processing systems such as biological evolution and
human cognition organize information used to govern the activities of natural entities.
When dealing with biologically secondary information, these systems can be specified
by five common principles that we propose underlie natural information processing
systems. The principles equate: (1) human long-term memory with a genome; (2)
learning from other humans with biological reproduction; (3) problem solving through
random generate and test with random mutation; (4) working memory when processing
novel information with the epigenetic system managing environmental information; (5)
long-term working memory with the epigenetic system managing genomic information.
These five principles provide an integrated perspective for the nature of human learning
and thought. They also have implications for the presentation of information.

Keywords: cognitive architecture, cognitive load theory, information processing
systems, long-term memory, working memory, random generate and test, evolution,
genetic system, epigenetic system, mutation.

¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Introduction

The suggestion that the development of human knowledge and biological
evolution by natural selection share a common underlying base has had a considerable
ancestry. That ancestry may stretch back to Darwin (1871), based on some
interpretations of his text. More recently, Campbell (1960), Dawkins (1976) and Popper
(1979) have clearly articulated an analogy between the processes of biological
evolution and knowledge development. Current work in this area emphasizes relations
between behavioral, cultural and biological evolution (e.g., Aunger, 2000; Boyd and
Richerson, 1985; Gintis, in press; Mesoudi, Whiten and Laland, in press). In a similar
vein, Siegler (1996) proposed that the acquisition of knowledge during cognitive
development was analogous to biological evolution. Furthermore, recent evidence that
cultural evolution has consequences for the biological evolution of species (Baumeister,
2004; Danchin, Giraldeau, Valone and Wagner, 2004) suggests that genetic and cultural
evolution may not only be analogous but interconnected, with each influencing the
other.
In much of this work, while human cognition is implicitly recognized as being
central, an explicit analysis of relevant aspects of human cognitive architecture is
missing. If knowledge development follows biological evolutionary principles, then the
mechanisms of human cognition and the structures that constitute human cognitive
architecture should incorporate the processes and functions of evolution by natural

---

Natural information processing systems

selection. In this paper we suggest that both human cognition when dealing with certain
categories of knowledge and evolution by natural selection provide examples of natural
information processing systems and that such systems can be specified by a series of
basic principles that detail the mechanisms of the system. We will begin by indicating
to which categories of knowledge the concept of natural information processing
systems can be applied.

Biologically primary and biologically secondary knowledge

Geary (2002, 2005, in press) has distinguished between biologically primary and
biologically secondary knowledge. Primary knowledge applies to categories of
information that we have evolved to acquire and use. Learning to listen to and speak a
native language, learning to interact socially with other humans and learning to use
general problem solving strategies that apply to a wide range of problems provide
examples of primary knowledge. Biologically secondary knowledge applies to
categories of knowledge that may have become culturally important relatively recently.
We have not had time to evolve specific mechanisms to deal with biologically
secondary information. Rather, we can adapt primary knowledge and its acquisition to
assist in processing secondary knowledge. Virtually all of the knowledge for which we
require schooling consists of secondary knowledge. Learning to read and write provides
a very clear example of secondary knowledge. We have not evolved to read and write
and so the manner in which reading and writing is learned differs markedly from the
manner in which listening and speaking develop.
Very large amounts of primary knowledge can be acquired easily, rapidly and
unconsciously. We do not require specific cultural institutions and procedures to
acquire such knowledge. It will be acquired automatically by all normal members of a
functioning society. In contrast, secondary knowledge must be explicitly taught and
learned via culturally organized procedures and institutions such as educational
institutions. Without appropriate institutions and procedures, secondary knowledge will
not be acquired by most members of a society. Thus, all normal members of a society
will learn to listen and speak simply as a consequence of being members of the society.
In contrast, very few members of a society learn to read and write unless the society has
organized deliberate procedures to facilitate such learning. For most individuals,
learning to read and write will not occur without deliberate cultural assistance because,
unlike listening and speaking, we have not evolved to automatically acquire these
skills.
The mechanisms by which biologically primary knowledge are acquired can be
assumed to be specific to the category of knowledge. Our ability to recognize faces or
learn the sounds of language are likely to be distinct. In contrast, our ability to acquire
biologically secondary knowledge must be general because by definition, we have not
evolved a capacity to acquire any particular category of that knowledge. We do have a
capacity consisting of procedures, possibly related to intelligence, to acquire general
secondary knowledge. Those procedures are required by any natural system that needs
to process a variety of categories of information and the remainder of this paper will be
concerned with the relevant processes.

Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 435 -

---

Natural information processing systems

Principles of natural information processing systems

Natural information processing systems can be found in nature. Like all
information processing systems, their function is to organize information that governs
the activity of entities incorporated by a system. Natural information processing
systems direct the activities of natural entities such as living organisms. There are
many ways of specifying the underlying logic of natural information processing
systems but in this paper we will focus on five basic principles (see Table 1) and
indicate how they apply to both human cognition and to evolution by natural selection.

Table 1. Natural information processing system principles

Principles
Cognitive case
Evolutionary case
Function

Information store
Long-term
Genome Store
information
principle
memory
for indefinite
periods

Borrowing and
Transfer
Transfer
Permit the rapid
reorganizing
information to
information to a
building of an
principle
long-term memory
genome
information store

Randomness as
Create novel ideas
Create novel
Create novel
genesis principle
genetic codes
information

Narrow limits of
Working memory
Epigenetic system
Input environmental
change principle
handling
information to the
environmental
information store
information

Environmental
Long-term
Epigenetic system
Use information
organizing and
working memory
handling genetic
stored in the
linking principle
information
information store


The information store principle


All natural information processing systems include a central store of
information that determines the bulk of activities of the system. Because the
environments in which natural information processing systems function are usually
complex, a very large store of information is required to handle the many conditions
faced. As a consequence, the size of the information store of natural information
processing systems is frequently too large and complex to measure in any more than
very approximate terms.
The contents of long-term memory provide the store of information for human
cognition and as a consequence, the bulk of human cognitive activity is directly
determined by long-term memory. The biologically primary knowledge associated with
what we see, hear, and think is governed by what we have previously learned and
stored in long-term memory and that primary knowledge, in turn, can be used to
acquire and store large amounts of biologically secondary knowledge. Initial evidence
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 436 -

---

Natural information processing systems

for the importance of long-term memory in human activities where it had previously
not been considered important came from de Groot’s (1946/1965) work on the game of
chess. Chess is a game of problem solving and it was easy to assume that long-term
memory played a minor role in successful chess playing. de Groot’s finding that chess
masters could much more accurately reproduce board configurations taken from real
games than week-end players, along with Chase and Simon’s (1973) replication and
demonstration that the difference disappeared using random board configurations,
altered our understanding of human cognition. No reliable differences between chess
experts and novices attributable to other factors have been found. Accordingly, it is
reasonable to assume that the only difference between expert and novice chess players
is due to experts having stored in long-term memory large numbers of chess board
configurations along with the best moves associated with each configuration. Novices
must attempt to work out the best move. Experts know the best move.
This result has been replicated in a variety of contexts using a variety of different
materials (e.g., Egan and Schwartz, 1979; Jeffries, Turner, Polson and Atwood, 1981;
Sweller and Cooper, 1985). The findings suggest that long-term memory does not just
provide the obvious function of permitting us to remember, it is central to all cognitive
activities including ones where memorization is not an obvious component. What a
chess grand master sees when he or she looks at a chess board configuration is different
to what a less able player sees just as what a person familiar with an office layout sees
when entering an office is likely to be vastly different to what a person who has grown
up in a forest and unfamiliar with offices sees when entering an office. The perceptual
differences are due to differences in the contents of long-term memory. Similarly,
problem solving moves that are obvious for a person familiar with a situation may be
impossible to contemplate for someone unfamiliar with that situation. Long-term
memory can both inform us of the characteristics of a situation and tell us how to deal
with it in the same way that a chess grand master’s long-term memory allows him or
her to recognize a board configuration and the most appropriate moves associated with
it.
If long-term memory has a central, critical function in cognition, it must be huge
to enable it to deal with the myriad of situations we face. Attempting to measure such
an entity is a formidable task, the more so since we have no appropriate metric. The
only attempt of which we are aware was conducted by Simon and Gilmartin (1973).
They limited their measure to the number of board configurations a chess grand master
is able to recognize and suggested that the number is between 50,000 and 100,000.
Since chess is only a part of life even for a chess grand master, the total capacity of
long-term memory is likely to be massive.
Evolution by natural selection is equally reliant on the information store
principle. In genetics, organized information determines the production of proteins and
resides in the genome. A genome is the total complement of an organism’s and/or
species’ genes and is central to genetic activity with evolutionary change focused on
genomic change. In cognition, if there is no change in long-term memory there has
been no learning. Similarly, if there is no change in a species’ genome, there has been
no evolution. Evolution means genomic change.
All genomes contain massive amounts of information and as was the case with
long-term memory, while there is no agreed procedure for measuring that information,
all conceivable measures indicate a very large information store (see, for example,
Portin, 2002 and Stotz and Griffiths, 2004, for discussions of techniques for measuring
the size of a genome). However they are measured, all genomes appear to require
thousands or even billions of units of information in order to allow life to survive and
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 437 -

---

Natural information processing systems

evolve. Thus a genome, like long-term memory, is a large information store that
governs complex activity, through very complex processes. That large information
store lies at the heart of natural information processing systems.

The borrowing and reorganizing principle

To fulfill its role, an information store must obtain large amounts of information.
In natural information processing systems the borrowing and reorganizing principle
provides the required mechanism. We suggest almost all of the semantic information
held in an individual’s long-term memory has been borrowed from the long-term
memory of other individuals (Boyd and Richerson, 1985). (Episodic memory is likely
to depend on individual perceptual experience.) Humans imitate other people, listen to
what they say and read what they write. These activities have the function of
transferring information from the long-term memory of one person to the long-term
memory of another in order to build the information store. Physical devices such as
books or electronic storage must frequently be used as intermediaries in this
transmission but all physically stored information initially came from an individual’s
long-term memory with the common intention of transfer to someone else’s long-term
memory.
While most information in long-term memory is borrowed, it is rarely, if ever,
borrowed without reorganization, either at the time it is borrowed or subsequently. The
borrowing and reorganizing process by which information is built in long-term memory
is constructive. Previous information is combined with new information to construct a
new representation with schema theory frequently used to describe the process (e.g.,
Chi, Glaser and Rees, 1982). A schema permits us to classify multiple elements of
information according to the manner in which we will use them. For example, we may
have a schema for a particular class of problems that permits us to classify the problem
elements according to the solution that is appropriate for that problem. Chess players
may classify board configurations according to the categories of moves appropriate for
each configuration and it is that knowledge that permits them to reproduce briefly seen
board configurations.
The process of combining new information with previous information has a
random component with random generation followed by tests of effectiveness
providing the mechanism. New information must be incorporated into previously
acquired schemas and the consequent new construct must be tested for effectiveness.
Because there is no way of determining whether the new construct is effective prior to
its construction, random generation followed by tests of effectiveness are required (c.f.
Simonton, 1999).
Evidence for the importance of the borrowing and reorganizing principle comes
from the worked example effect. In the many experiments demonstrating this effect,
learners who were presented worked examples to study rather than the equivalent
problems to solve were better able to solve subsequent test problems (Carroll, 1994;
Cooper and Sweller, 1987; Miller, Lehman and Koedinger, 1999; Paas, 1992; Paas and
van Gog, 2006; Paas and van Merriënboer, 1994; Pillay, 1994; Quilici and Mayer,
1996; Reisslein, Atkinson, Seeling and Reisslein, 2006; Sweller and Cooper, 1985;
Trafton and Reiser, 1993; van Gog, Paas and van Merriënboer, 2006). Learners
presented mathematics or science worked examples to study, in effect, were borrowing
problem solutions from other people while learners presented with problems to solve
were devising their own solutions. Worked examples, by indicating an appropriate
solution, reduce or eliminate random problem solving attempts. The more substantial
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 438 -

---

Natural information processing systems

learning following the study of worked examples demonstrates the effectiveness of the
borrowing and reorganizing principle.
The instructional use of worked examples when learning to solve problems is a
form of imitation. The recent discovery of a mirror neuron system in both monkeys
(Gallese, Fadiga, Fogassi, and Rizzolatti, 1996) and humans (Grafton, Arbib, Fagiga,
and Rizzolatti, 1996) for motor action provides physiological evidence for the
importance of imitation in human cognition. The mirror neuron system is an
observation-execution matching system that fires when an individual either observes or
executes an action. The fact that the system activates when observing others as well as
when acting oneself indicates the importance to human cognition of observing the
actions of others.
The mirror neuron system is involved in imitation. Iacoboni, Woods, Brass, et al.
(1999) tested the consequences of asking people to make finger movements using
several different sets of instructions. In one case, they were asked to imitate a hand in
which a finger was lifted while in other cases a cross appeared above the finger that
was to be lifted or a signal associated with lifting a particular finger was presented. The
authors found that the mirror neuron system became active under all three conditions
but a larger signal intensity was obtained for the imitative than the two non-imitative
conditions. Iacoboni et al. also found that the system became active when people were
asked to merely observe rather than observe and act under the three conditions. Further
work has found that the mirror neuron system not only fires when an action is observed
or signaled, it also fires when people listen to sentences describing actions (Tettamanti,
Buccino, Saccuman et al., 2005).
We can conclude that imitating other people’s actions either directly seen or
inferred from speech or signals is an important method of obtaining information –
sufficiently important for us to have evolved physiological mechanisms specifically to
handle imitation. While our evolved tendency to imitate is biologically primary, there is
no reason to suppose that the knowledge acquired is necessarily primary. The same
evolved tendency to imitate is likely to apply to biologically secondary information.
Imitation by humans may be universal whether it involves a simple physical action
probably based on primary knowledge or a complex mathematical procedure based on
secondary knowledge. We are physiologically organized to use the borrowing and
reorganizing principle.
The borrowing and reorganizing principle is deeply entrenched in biological
evolution. When one generation reproduces the next, the new generation borrows
genetic material from the parent generation. Asexual and sexual reproduction provide
two mechanisms of information transmission with substantially different information
processing characteristics.
During asexual reproduction, a single individual passes a copy of all of its
genome to its offspring. That genome can be copied and repeatedly passed on to new
individuals during a process that seems to have no equivalent in human cognition
because of the limited role reorganization plays in asexual reproduction. Theoretically,
there is no variation in the offspring produced by asexual reproduction with all the
information available to one generation also available to the next. In fact, two sources
of variation can occur. The first, mutation, will be discussed under the next principle.
The second provides several examples of the borrowing and reorganizing principle.
Variability, through the exchange of genetic information, can occur in simple, asexually
reproducing organisms like bacteria and viruses. There are three mechanisms that
transfer genes between such individuals, with the details of the processes different from
those involved in sexual reproduction (Redfield, 2001). 1. Bacteria can directly transfer
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 439 -

---

Natural information processing systems

genetic material between two individual cells that join temporarily (a process which
may be considered as the bacterial version of sexual reproduction). One cell donates its
DNA and the other receives the genes. 2. Bacteria can alter their genomes by absorbing
DNA from the environment and incorporating pieces into their own chromosomes. The
cell is now a recombinant: its chromosome containing DNA obtained from two
different cells 3. Viruses that infect bacteria can transfer bacterial genes from one host
cell to another. The process results from by-products generated during the infection by
the virus.
In all these situations, information is borrowed and reorganized. As will be noted
below with respect to sexual reproduction, each of these processes is likely to have a
random component. Which genetic material is borrowed by these mechanisms and
when it is borrowed is likely to be random. If the new material is adaptive, it will
persist in subsequent generations. The techniques of reorganization described above
occur in asexually reproducing organisms. Of course, asexual reproduction typically
only involves borrowing by direct copying without reorganization.
While asexual reproduction does not usually reorganize information, sexual
reproduction always and necessarily involves reorganization of information. During
sexual reproduction, two cells are needed to reproduce offspring with each parent cell
providing genetic material. This procedure has three basic and closely related
information processing consequences. First, each new individual is a “construction” of
its parents’ genetic material rather than a replication. Second, the process of sexual
reproduction, by its very nature of fusing genetic material from two individuals,
eliminates the possibility of exact reproduction and thus, unlike asexual reproduction,
the material borrowed by the offspring always varies from the information possessed by
the parents. In other words, reorganization is an essential part of the process. Third, not
only are offspring genetically different from their parents, they are almost always
different from each other (with the exception of identical siblings). Sexual reproduction
is a technique for increasing the diversity of genetic information in a species with
incremental alterations to the genetic information passed from generation to generation
and it is that increased diversity that provides sexual reproduction with its advantage
over asexual reproduction (e.g., Hamilton and Zuk, 1982).
In contrast to asexual reproduction, the evolutionary more recent sexual
reproduction is closely analogous to the constructive processes of schema formation.
As is the case for human cognition, there are unavoidable random components to sexual
reproduction and it is those components that provide the variability generated by sexual
reproduction. There are two main phases during sexual reproduction: the formation of
male and female sex cells followed by the union of a male and a female sex cell.
During both of these stages, genetic material is randomly reorganized, resulting in
limitless variation, such that offspring will always be unique, with the partial exception
of identical siblings.
There are three other important genetic processes that use the borrowing and
reorganizing principle.
1. Most human genes contain two types of segments: one type codes for
polypeptides (protein sub-units), the other has no protein coding function. Following
the first step of gene expression (when the DNA code is transcribed into RNA),
splicing of the original gene code occurs by excising the non-coding sections and then
joining the coding sections in a way that will ultimately form the template for a
particular protein. However, the coding RNA sequences can be spliced in alternative
ways, thus rearranging the order of the coding sections and thereby coding for a
different protein. As a consequence of splicing, one gene can produce more than one
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 440 -

---

Natural information processing systems

protein. The alternate arrangement or splicing of codes becomes important when the
same information needs to have a different function at different times, such as during
the development of an organism or when the same information must be used to provide
very high levels of diversity (Modrek and Lee, 2002). Thus, by the process of
alternative splicing, under new environmental conditions and new requirements of the
organism, new proteins with new, potentially useful, functions can be formed. It has
been suggested that the process of alternative splicing is associated with evolutionary
change (Modrek and Lee, 2003). The process is relevant to the borrowing and
reorganizing principle because, as is the case with sexual reproduction, new
information is not created directly. Rather, previously created information is borrowed
and rearranged.
2. Viruses may provide one of the techniques by which non-coding sequences are
inserted into DNA. Viruses can reproduce only inside other cells, using the genetic
machinery of the host cells. Viruses inject their own genetic material into the host cell
and copies of the viruses are made using the host DNA. Over time, some parts of the
viral DNA may remain in the nuclei of some host cells. It has been suggested that these
remnant pieces of DNA, which previously had coding capabilities, may have randomly
inserted into the host DNA and have lost their coding potential. They now make up the
non-coding segments (i.e. segments that do not code for protein) of the nuclear DNA
molecules and may have regulatory functions (Rogozin, Babenko, Fedorova, et al.,
2003; Turner, 2001; Weinzierl, 1999). This process also provides an example of the
borrowing and reorganizing principle.
3. Another method of rearranging stretches of DNA occurs when sections of the
DNA move (mobile genetic / transposing elements) from one location to another within
the genome and so alter the output of many genes. This movement is more likely to
occur in active regions of DNA chains (Jablonka and Lamb, 1995; 2005) and the
activation of transposable elements has been shown to occur in stressful circumstances
(McClintock, 1984). Again, the new information resulting from the rearrangement has
been built by “borrowing” previously created information. It should be noted that both
alternative splicing and mobile genetic elements are controlled by the epigenetic
system, discussed below.
Reorganizing previously organized information in these ways does not guarantee,
of course, that the newly organized information will be adaptive. There is an aspect of
random generate and test (see next principle) in both alternate splicing and mobile
genetic elements. For the moment, the similarity of these genetic mixing procedures
and the manner in which the human cognitive system will combine previously acquired
information to generate new information needs to be noted. For example, whenever a
problem is solved by analogy (e.g., Gick and Holyoak, 1980, 1983), information from
the source analogue is combined with information from the target problem to produce a
new problem solution. That attempted solution may or may not provide an actual
solution and so the analogy needs to be tested for effectiveness. Whenever knowledge
in one area is combined with knowledge in another area, new information is produced
that is equivalent to gene splicing and mobile genetic elements.
The borrowing and reorganizing principle is the major mechanism by which
natural information processing systems provide individuals with large information
stores, either cognitive or genetic. The principle permits the rapid acquisition of huge
amounts of information that could not otherwise be acquired. Furthermore, both
biological evolution and human cognition are structured to reorganize that information
at the time it is borrowed, test the effectiveness of the resultant reorganization and
retain or jettison it depending on the outcome of the test. While the vast bulk of
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 441 -

---

Natural information processing systems

cognitive and genetic information held by an individual is acquired via the borrowing
and reorganizing principle, it is not the only source of information. New information is
created via the next principle, the randomness as genesis principle.

The randomness as genesis principle


Strictly speaking, the borrowing and reorganizing principle reorganizes rather
than creates new information although, of course, reorganization does involve
creativity. Nevertheless, a distinction does need to be made between reorganizing
previously created information and an original act of creation that results in new
information. A procedure for creating new knowledge is required because for a variety
of reasons, useful, previously organized information obtainable via the borrowing and
reorganizing principle may not be available. Consequently, natural information
processing systems need distinct procedures for, on the one hand, borrowing and
reorganizing previously generated information and for creating new information on the
other hand.
Humans create new information during problem solving. The problem solving
strategy that has undergone the most detailed study is means-ends analysis (Newell and
Simon, 1972). This strategy was extensively analyzed using computer modeling during
the 1970s and 80s (e.g., Sweller, 1988). In essence, the strategy requires a problem
solver to repeatedly consider a current problem state, consider the goal state, extract
differences between them and find a problem solving operator that can reduce those
differences. Successful problem solutions can result in the creation of new knowledge
that can be stored in long-term memory for subsequent use.
A close analysis of a means-ends strategy or, indeed, any problem solving
strategy intended to discover new solution procedures will reveal that random
generation followed by tests of effectiveness is central to the strategy. Consider a
problem solver who has extracted differences between a current problem state and the
goal state using a means-ends strategy. The next step is to find a problem solving
operator that will reduce those differences. That process is straightforward if the
problem solver either has knowledge in long-term memory indicating which problem
solving operators might be used to reduce the differences or has access to knowledge in
someone else’s long-term memory. For example, competent elementary algebra
problem solvers will know that if faced with the problem, (a + b)/c = d, solve for a,
that multiplying both sides by the denominator c will reduce differences between the
current and desired goal states. Prior knowledge can be used to generate this move but
while the relevant knowledge may be strengthened through automation (Kotovsky,
Hayes and Simon, 1985), new knowledge is not generated.
In contrast, consider a problem solver who has just learned the relevant rules
(problem solving operators) of algebra. The problem solver does not have schematic
knowledge in long-term memory indicating which moves are relevant to solving this
problem. Under these circumstances, failing the receipt of information from others, we
suggest the only viable strategy is to randomly generate a legal move and test it for
effectiveness by observing whether the move has the desired effect of reducing
differences between the given and goal states. Failing knowledge in long-term memory,
there is no procedure available for determining the effects of a possible move prior to
selecting that move. Accordingly, random selection is the only procedure available. In
order to determine whether a move will reduce differences between the given and goal
states a problem solver must randomly select it and either mentally or physically test it
for effectiveness.
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 442 -

---

Natural information processing systems

If the randomly selected move has the effect of reducing differences between the
given and goal states, both the move and those states may be remembered on
subsequent occasions obviating the need for random selection. In this manner, new
information has been generated that can become part of the information store.
Furthermore, this process provides the genesis of all new knowledge. The knowledge
that we acquire from others via the borrowing and reorganizing principle had to be
generated in the first instance by the randomness as genesis principle. Without this
principle, basic, new knowledge cannot be generated.
It must be emphasized that random move generation during problem solving will
only be used in the absence of any knowledge for ranking moves. If, for example, in the
absence of complete knowledge, sufficient knowledge is available to rank a series of
plausible alternatives, that incomplete knowledge will be used to generate moves.
Nevertheless, under at least some circumstances, knowledge may be available to
generate two or more possible moves that cannot be ranked on the basis of that
knowledge. Under these circumstances, there is no available mechanism to choose a
move to test other than random choice. Evidence for random choice under these
circumstances comes from errors during problem solving. When faced with unfamiliar
problems, most moves are likely to result in dead-ends, a result that may be difficult to
explain under conditions other than random generation.
Evolution by natural selection uses a structurally identical procedure to human
cognition to generate new information. New information is created by mutation
(changes in DNA) using a similar procedure to humans solving a novel problem. As is
the case during problem solving, because mutations are random, most are not adaptive
and lead to “dead-ends”. While random generation is central, because most randomly
generated mutations are not adaptive, random generation must be followed by tests of
effectiveness.
The “problem” faced by all living organisms is survival and reproduction in a
particular environment. Survival and successful reproduction provide evidence of
effectiveness. As is the case with human cognition, there is no a priori system available
to determine whether a possible mutation is likely to be useful. That determination only
can be made after the event with successful mutations leading to increased offspring
and unsuccessful mutations leading to decreased offspring. Furthermore, this process of
mutation is the genesis of all biological variation. During asexual reproduction, apart
from the probably rare borrowing of information from other cells as indicated
previously, there can be no other source of variation. With respect to sexual
reproduction, all the variation between the male and female genetic material (DNA),
can be sourced back to a series of mutations. Without those mutations, the male and
female DNA would be identical, resulting in no benefits of sexual over asexual
reproduction. In other words, the advantages of the constructive processes that are
integral to the borrowing and reorganizing principle rely on a series of prior mutations
that occur according to the randomness as genesis principle.
For a mutation to be inherited from one generation to the next in sexually
reproducing organisms, the change must be within the sex cells. If the change is in a
normal body cell, then only a subgroup of cells in the individual will be affected but the
mutation will not be passed on to the offspring organisms. This modification of sex
cells is the basis of evolution. There are different kinds of mutations and their impacts
vary. As might be expected of a random process, many mutations are deleterious, some
are adaptive and some are neutral with selection pressures having no net effect.
While mutations are normally considered to be random, Jablonka and Lamb
(2005) have suggested that in some situations mutations may occur in a non-random
Evolutionary Psychology – ISSN 1474-7049 – Volume 4. 2006. - 443 -

---

Document Outline

• ??

---