Nonlinear Dynamics, Psychology, and Life Sciences, Vol. 15, No. 2, pp. 145-173.

© 2011 Society for Chaos Theory in Psychology & Life Sciences

Nonlinear Dynamics and Neo-Piagetian Theories in

Problem Solving: Perspectives on a New Epistemology
and Theory Development
Dimitrios Stamovlasis1, Aristotle University of Thessaloniki, Greece

Abstract: In this study, an attempt is made to integrate Nonlinear Dynamical

Systems theory and neo-Piagetian theories applied to creative mental processes,
such as problem solving. A catastrophe theory model is proposed, which
implements three neo-Piagetian constructs as controls: the functional M-
capacity as asymmetry and logical thinking and the degree of field dependence-
independence as bifurcation. Data from achievement scores of students in tenth
grade physics were analyzed using dynamic difference equations and statistical
regression techniques. The cusp catastrophe model proved superior comparing
to the pre-post linear counterpart and demonstrated nonlinearity at the
behavioral level. The nonlinear phenomenology, such as hysteresis effects and
bifurcation, is explained by an analysis, which provides a causal interpretation
via the mathematical theory of self-organization and thus building bridges
between NDS-theory concepts and neo-Piagetian theories. The contribution to
theory building is made, by also addressing the emerging philosophical, -
ontological and epistemological- questions about the processes of problem
solving and creativity.
Key Words: nonlinear dynamics, catastrophe theory, creativity, problem solving,
neo-Piagetian theories, M-capacity, logical thinking, field dependence-
independence, science education
INTRODUCTION
Problem solving, innovation and creativity are of a major concern in
psychology, education, management and other social sciences. This justifies the
large amount of research work published for the last decades in these fields,
which have fostered various theoretical frameworks in order to state and test hy-
potheses and explain empirical data.
In the field of science education and specifically in problem solving, re-
searchers have fostered neo-Piagetian theories (Pascual-Leone, 1970) and infor-
mation processing models (Baddeley, 1986) in order to understand and explain
students’ performance. Both frameworks account for the limited human channel
1
Correspondence address: Dr. Dimitrios Stamovlasis, Aristotle University of
Thessaloniki, Faculty of Philosophy, Dept. of Philosophy and Education, 54 124
Thessaloniki, Greece. E-mail: [email protected]
145
146 NDPLS, 15(2), Stamovlasis
capacity, by implementing the concepts of Mental (M-) capacity and working
memory capacity respectively, since subjects either when solving problems or
learning material, they have to hold and process information. A simple model,
then, states that, as the problem increases in complexity and if the human chan-
nel capacity has a final limit, the decrease of achievement may be rapid after the
limit has been reached (Pascual-Leone, 1969; Scandarmalia, 1977). This is
known as the overload hypothesis, which has been demonstrated by an inverse
S-shaped curve, which is the graph of subjects’ performance as a function of the
demand of a problem. However, from a methodological and statistical point of
view, the human-channel capacity-overload effect, which implies discontinue-
ties, was not expressed by means of an appropriate nonlinear mathematical
model. In addition, it is well established that a number of other individual differ-
ences, such as logical thinking, field dependence-independence, convergent-
divergent thinking or mobility-fixity dimension, are of great importance in prob-
lem solving processes (Johnstone & El-Banna, 1986; Johnstone & Al-Naeme,
1991; Niaz, 1988, 1989; Niaz, de Nunez, & de Pineda, 2000; Tsaparlis, 2005;
Tsaparlis & Angelopoulos, 2000; Danili & Reid, 2006; Stamovlasis, Tsitsipis &
Papageorgiou, 2010). The role of the above variables has been demonstrated not
only in problem solving, but also in students’ understanding science, that is, the
role of the above cognitive variables is important during learning –and teaching
(Johnstone & Selepeg, 2000; Stamovlasis & Tsaparlis, 2003; Tsitsipis,
Stamovlasis & Papageorgiou, 2010). Learning or conceptual understanding as a
‘stage of mind’ is the product of a learning process, where interplay among these
variables is involved. This area of investigation concerns mental processes and
could be considered as the interface between educational science and psycholo-
gy. It borrows theories from psychological sciences in order to explain
phenomena taking place in educational environments, such as learning, problem
solving, novelty, insight and creativity. However even thought this area deals
with phenomena evolving in time, it has very little been affected by the
epistemological, methodological and theoretical advancements in psychological
and brain sciences.
Thus, the above field has two main disadvantages. First, it fostered
machine analogies for cognition and it has insisted on this view for quite long
time without having been influenced even by the well-known neural
development theories and connectionism (Edelman, 1987, 1989). Thus, learning
paradoxes and philosophical questions, such as, ‘homunculus’ problems still
remain unanswered. The second disadvantage is the implementation of the
general linear model (GLM). While more variables were found to play an
important role in problem solving, the GLM even if it includes all potential
predictors, it still leaves a large portion of the variance unexplained. Moreover,
among other methodological problems, GLM does not provide functional roles
for the variables involved and most importantly, it is epistemologically
incompatible with the system and inadequate to access information or
knowledge about the process under investigation (Stamovlasis, 2010).
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 147
Cognitive Variables
In this section a brief description of the variables involved in the
present research is provided:
M-capacity
It is a neo-Piagetian construct accounting for the limited human chan-
nel. It refers to functional capacity being a portion of the structural capacity used
by a subject at any particular moment of his or her cognitive activity (analogous
distinction is the potential versus usable working memory). M-capacity consti-
tutes a discrete random variable and can be influenced by a multiplicity of fact-
ors (Pascual-Leone, 1969, 1970; Witkin Dyk, Paterson, Goodenough, &Karp,
1962). The theory emphasizes the importance of a match between the subject’s
M-capacity and the mental demand of a task (see next section on Pascual-
Leone’s Theory).
Logical Thinking
Logical thinking is a Piagetian concept and refers to the ability of the
subject to use concrete- and formal-operational reasoning and is also referred as
the developmental level (Lawson, 1978, 1985, 1993). Logical thinking was as-
sessed by the Lawson test, a pencil-paper test of formal reasoning (Lawson,
1978), which includes the following reasoning modes: Proportional, combina-
tional and probabilistic reasoning as well as reasoning related to the isolation
and control of variables, conservation of weight and displaced volume. In
science education, but not merely, logical thinking has been found to play a
major role in students’ performance.
Field Dependence-Independence
Field dependence-independence is a cognitive style associated with
one’s ability to disembed relevant information from complex and potentially
confusing contexts (Witkin et al., 1977). Psychological investigations have
shown that some people are dominated by any strong frame of reference or pat-
tern in a stimulus field to such an extent that they have trouble in perceiving
elements that cut across the pattern. Those who inefficiently separate an item
from its context and readily accept the dominating field (context) are
characterized as field-dependent. On the contrary, those who can easily
penetrate an organized perceptual field and separate an item from its context are
characterized as field-independent Disembedding ability is related to structural
ability, so that field-independent individuals are more able to deal with ill-
structured tasks than field-dependent individuals (Witkin et al., 1962; Witkin &
Goodenough, 1981). There are many identifying tasks in the literature that are
described as measuring perceptual style, which could be found in lucid reviews
(Milne1 & Szczerbinski, 2009). The two above qualities, field dependence-field
independence, are not regarded as two distinct categories, but a continuum
between them there exists, leaving space for an intermediate ability, where
148 NDPLS, 15(2), Stamovlasis
subjects are classified as field-intermediate. In problem solving literature, field
dependence- independence has been related to the information processing
models as a moderator variable. Field-dependent subjects appear to possess
lower information processing ability, since part of their capacity is being used to
process irrelevant information. Research literature on cognitive styles is huge
and it has consolidated a strong correlation with learning outcomes, problem sol-
ving abilities and academic performance in various disciplines, such as, lan-
guage, mathematics, natural sciences, computer sciences, social sciences and art
(Tinajero & Paramo, 1998). Moreover, an association of cognitive style and cre-
ativity has also been reported (Miller, 2007).
The Complexity and the Nature of the Problem
The effect of the above psychometric variables on subjects’ perfor-
mance depends also on the complexity and the nature of the problem. The com-
plexity of a problem in science education is described by (a) the mental demand
and (b) the logical structure of the problem. One definition of demand is “the
maximum number of thought steps and processes which have to be activated by
the least able, but ultimately successful candidate in the light of what has been
taught” (Johnstone & El-Banna, 1986). The logical structure of the problem is
associated with the number of logical schemata, which the solver has to retrieve
from his long-term memory in order to solve the problem. When the complexity
of the problem is low, the effect of the above psychometric variables is small,
while in very demanding mental tasks the effect becomes more pronounced.
Concerning the nature of the problem, all cognitive tasks can be classi-
fied into two broad categories (Nicolis, 1986): In the first category are the de-
ductive-type cognitive tasks whose solutions always exist. The solutions of these
problems are unique and are implicitly contained in the initially given data set or
premises, provided that these premises form a complete and self-consisted set.
During such processes no new information is produced.
In the second category are the complex cognitive tasks which are in-
ductive in nature and whose solution may not be unique or may not exist at all.
The information required for the solution is not given exclusively in the initial
data, but is generated by the evolution of problem solving process (Nicolis,
1991).
In science and mathematics education, teaching problem solving fosters
an analogous distinction between cognitive tasks called exercises and (“real”)
problems. Exercises are mental tasks, where the subject applies a well-known
procedure, which has usually been practised. Reaching the final outcome-solu-
tion, it is credited to the student’s ability to execute successfully the learned
algorithm. It is imperative to say that in these processes the subject may even be
ignorant of the strategy followed or of how to turn the implicit into explicit.
On the other hand, “real” problems or complex problems, which are not
algorithmic, faced by a novice, mimic or simulate inductive-type problems, in
the sense that for him or her the outcome is not obviously nested in the initial
conditions or given data, and the solution is not guaranteed by the students’
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 149
previous repertoire. The strategy to be followed is not known or it might not be
unique, and the outcome is in part generated by the evolution of the mental
process. We must emphasize that in a school context a cognitive task can be an
exercise or “real” problem depending on the subject’s expertise and on what has
been taught. A task could be a “real” problem for a student having a certain
repertoire, while the same task is an exercise for another student (Niaz, 1995).
The mental processes corresponding to the above two different in na-
ture tasks are totally different when they are seen as patterns unfolding in time.
This is a basic point for the thesis of the present paper. While the execution of
an algorithm is a cyclical or linear process, the mental processes in “real” pro-
blem solving, which are processes of creativity and innovation, exhibit nonlinear
dynamic characteristics.
Neo-Piagetian Theories: Pascual-Leone’s Theory of Constructive Operators
The Theory of Constructive Operators (TCO) was originally proposed
by Pascual-Leone (1970, 1987, Pascual-Leone et al., 1978) as an account of in-
dividual differences in performance on cognitive tasks. It is often referred to as
‘neo-Piagetian’ theory in that it offers an account of cognitive performance of
children at different ages, linking these age-related differences to different
stages of cognitive (Piagetian) development. A subject is viewed as a relatively
autonomous psychological system, which can be described by means of the
following three components: (a) the repertoire H of behavioral units or schemes;
(b) a central computing space M (M-operator) in which the information proces-
sed by the set H* of activated schemes is transformed or integrated into a novel
behavior; (c) a number of organizational laws, such as learning laws, field
organization laws, etc. (Pascual-Leone, 1970).
According to this theory, the information processing ability of a sub-
ject, that is, his or her mental capacity (or M-capacity), represents “…a reserve
of mental energy that can be allocated to raise the activation weight of task-rele-
vant schemes. This reserve of M-energy may be measured by the maximum
number of discrete ‘chunks’ of information or different schemes that M can con-
trol or integrate in a single act of centration…” (Pascual-Leone et al., 1978).
The term scheme refers to a well-learned procedure that can be ‘activa-
ted’ and applied in order to accomplish a task. The notion of centration refers to
the coordinated use of these various well-learned procedures (or schemes) for
the performance of a given task. Coordination of the schemes is the responsibili-
ty of a variety of constructive operators (CO), each of which performs a specific
function: the M-operator deals primarily with mental capacity; the C-operator
with content knowledge, the L-operator with logical operations such as conser-
vation and formal logic; the F-operator with field dependence-independence;
and finally, the I-operator is responsible for the deactivation of task irrelevant
schemes.
A distinction should be made between subjects’ maximum structural
capacity (Ms) and functional capacity (Mf), the latter being a portion of Ms used
by a subject at any particular moment of his or her cognitive activity. Mf
150 NDPLS, 15(2), Stamovlasis
constitutes a discrete random variable and can be influenced by a multiplicity of
factors (Witkin et al., 1962; Pascual-Leone, 1969, 1970). An analogous distinc-
tion can be made with working memory: potential working memory versus us-
able working memory (Johnstone & Al-Naeme, 1991). Mf oscillates between
zero and Ms.
Pascual-Leone’s model emphasizes the importance of a match between
the subject’s M-space and the mental demand of a task. This demand refers to
the maximum number of schemes, which the subject must activate simultane-
ously in the course of successfully executing a task, and is determined by means
of task analysis that identifies how many schemes would be required to solve the
problem (Case, 1978; Pascual-Leone & Goodman, 1979). These schemes are ap-
parently the thought steps and processes in the definition of mental demand
given at the introduction. The implication is that it is possible for the subject to
fail to perform a task because the mental demand of the task exceeds the
subject’s M-capacity. Significant correlations have been reported between M-
capacity and mental demand in numerous studies (e.g. Niaz & Lawson, 1985;
Niaz, 1996). Even small changes in mental demand can lead to M-capacity over-
load; furthermore, the manipulation of the logical structure (that is the degree to
which the problem requires formal operational reasoning) could also lead to
significant changes in the students’ performance (Niaz & Robinson, 1992).
A basic advantage of TCO is that it provides measurable variables that
can operationalize the theoretical concepts (CO). Furthermore, a theoretical de-
velopment would be attained if the operations of CO were viewed as processes,
evolving in a complex interwoven way, rather than as executive components in a
machine analogy paradigm. This will allow a bridge with NDS and especially
with self-organization theory.
According to neo-Piagetian TCO in a cognitive task execution, the
mind activates certain operative schemes, which are responsible for the transfor-
mation and coordination of the preexisting or the new information (or schemes
or knowledge) with the action of constructive operators each of which performs
a specific class of function (Pascual-Leone, 1970, 1974, 1989). At the level of
complexity at which psychology and science education research studies the cog-
nitive process, one could recognize a number of mental sub-processes, which are
related to variables operationalizing the constructive operators. Such variables
are the information processing capacity, logical thinking, field-dependence-inde-
pendence or convergent-divergent thinking.
In a mental-task execution, the constructive operators are successively
activated in order for the mind to proceed and reach a final outcome, a solution,
a novelty or creation. The pattern of these successive activations is the core issue
of the theory developed in the present thesis. This pattern is determined by the
nature of the mental task. An average contribution of each operator to a mental
task is reflected on the measured statistical effect of the corresponding variable
as portion of the variance explained by GLM. However, the GLM when imple-
mented does not reveal any clue about those patterns. These patterns are a prime
concern for NDS.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 151
Piagetian Theories and Nonlinear Dynamics
Piagetian theory has been related to nonlinearity through catastrophe
theory (CT), which was connected to Piagetian stagewise development
(Molenaar & Oppenheimer, 1985; Lewis, 1994). CT models in science involve
dissipating systems or potential-minimizing systems. Such models ignore the
very large number of internal variables, and they constrain the description of the
local observed behavior by a small number of control parameters (Feraro, 1978);
Poston & Stewart, 1978; Gilmore, 1981; Castrigiano, & Hayes, 2004). This
makes CT of unique importance to the analysis of complex biological or psycho-
logical possesses. One CT theoretical model has been proposed for the balancing
between assimilation and accommodation, which was sought as introducing the
dynamics that could lead to sudden jumps to a new stage in development (Saari
(1977), while a cusp catastrophe model has been constructed for the transition
from preoperational to concrete operational thought (Preece, 1980).
An important aspect is that the concept of equilibration (Piaget, 1995)
as applied to Piagetian and neo-Piagetian theories can be covered by the defini-
tion of equilibrium in Catastrophe Theory, which is expressed mathematically
by setting the first derivative of the dynamic system equation to zero. A method-
ological advancement in favor to CT was the development of a statistical theory
applicable to empirical data (Cobb & Zacks, 1985). Moreover, statistical me-
thods and techniques were developed for empirically testing CT models (Cobb
et al., 1983; Cobb & Watson, 1980; Grasman, et al., 2009; Guastello, 1982,
1995; Oliva, Desarbo, Day, & Jedidi, 1987).
There is a considerable number of CT applications or theoretical en-
deavors in psychological science, a review, however, is out of the present scope,
but some characteristic ones include: parameterization of learning curves (Baker
& Frey, 1980), motivation and academic performance (Guastello, 1987), pro-
blem solving and creativity (Boles, 1990), reasoning (van der Maas, Jansen, &
Raijmakers, 2004), language acquisition (Bernardez, 1995), human development
(Lewis, 1994, 2000; Molenaar & Raijmakers, 2000; Thelen & Smith, 1994; Van
der Maas, & Hopkins, 1998; van Geert, 1991, 2003), -to mention a few.
Catastrophe models for stage transition by employing control variables from
neo-Piagetian theories, such as theories of Case (1985) and Pascual-Leone
(1970) were also proposed (van der Maas & Molenaar, 1992). Note that con-
cerning the role of neo-Piagetian variables (information processing capacity,
logical thinking, field dependence-independence, or convergent-divergent think-
ing), it has been well-established in science education, however, within the line-
ar framework. Lately a shift towards the nonlinear regime was made by testing
cusp models using dynamic difference equations and statistical regression tech-
niques. It was shown that, in some cases, cusp catastrophe models are superior
to the linear counterparts and that bifurcations and hysteresis effects (Figs. 1, 2
& 3) are observed in students’ performance when the demand of the problem in-
creases or the algorithmic character of the problem decreases (Stamovlasis,
2005, 2006, 2008).
152 NDPLS, 15(2), Stamovvlasis

Fig. 1.. A three-dimeensional repre

esentation of tthe cup catas trophe respon
nse
surface
e for achievement in science education prooblem solving. The arrow in tthe
Logicall Thinking dime
ension denotes
s lower values.

Fig. 2. Bifurcation in problem

p solving
g.

Fig. 3. An example off hysteresis effe

ect in problem solving achievvement.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 153
In the present paper further empirical evidence is provided for
nonlinear effects at the behavioral level. In addition, a contribution is made
towards the theoretical development of the neo-Piagetian framework applied to
problem solving, which could be generalized to analogous qualitative-change
processes, such as creativity, insight, novelty and the genesis of an artistic in-
spiration or a scientific conception or even a scientific revolution.
Problem Solving as a Nonlinear Process
Having examined the different types of problem solving, it is essential
at this point to show where the nonlinear dynamics are relevant. The following
paragraphs provide a verbal description of the nonlinear characteristics of men-
tal processes, which could be associated with creativity.
Consider the deductive-type cognitive tasks or the execution of algo-
rithmic type of problems, whose solutions always exist: The outcomes of these
problems are unique and are implicitly contained in the initially given data. In
such cognitive tasks the mind proceeds through regressive steps on a sequence,
which finally converges and gives the solution. Next consider complex cognitive
tasks, which are inductive in nature and whose solution may not be unique or
may not exist at all: The outcome in this type of problem solving is not guaran-
teed from the subject’s previous experience or his or her existing repertoire. The
case is analogous for novices’ minds facing a “real” problem. For this type of
complex problems, there is no unique path to follow and each step is determined
by the previous steps. The information required for the solution is not hidden ex-
clusively in the initial data, but it is generated by the evolution of an iterative
and recursive process. These are dynamical processes, which produce informa-
tion for the system and whose outcome can be understood in an evolutionary
context (Nicolis, 1991). These are the creative processes.
In a cognitive task execution, according to neo-Piagetian theories, the
mind activates certain operative schemes, which are responsible for the transfor-
mation and coordination of the preexisting or the new information (or schemes
or knowledge) with the action of constructive operators (Pascual-Leone, 1970,
1974, 1989). The mental sub-processes (operationalized by variables, such as
those implemented as controls in the proposed cusp model) are the state para-
meters of the evolving system, which is characterized by the pattern of these
successive activations. Thus, in a mental task execution, a mind might proceed
as follows: inputs data (I), retrieves information from long term-memory (R),
processes information (M), separates “signal” from “noise” (if any) (F), applies
formal reasoning (L), processes information (M) again and so on. Symbolically:
IIRMMLFRMILLFFMFLCRLIICLLMMRFFM…
In executing a familiar algorithm, like in the deductive type of mental
tasks, the successive activations of various schemes and operators follow a step-
by-step linear or cyclical procedure. The sequence of the above subprocesses is
predetermined. The solution is nested in the algorithm and in the initial
conditions or data.
154 NDPLS, 15(2), Stamovlasis
In a “real” problem situation, like in the inductive type of problem - a
creative process - the sequence of the above steps is not predetermined. The
mind might activate successively the above subprocesses, in a pattern that is
neither linear nor cyclical and might be seemingly random. This pattern emerges
from an interwoven interaction among mental operators, and it might exhibit
self-similarity and fractal structure. The successive mental functions are selected
by a random choice (microscopic fluctuations) with no preexisting scenario. The
outcome of each subsequent pseudo-random step is integrated with preexisting
schemata by chunking mechanisms (Newell, 1990) as temporal ‘gain’ or ‘loss’
and consequently might feedback the next steps. In such dynamical process,
both convergence and divergence are present and the interplay between them,
analogously to the role of competence versus cooperation in perceptual motor
learning (Kelso, 1995), determines the emergence of the final outcome. Given
that each step is determined by the previous steps, the process becomes history
dependent, in Prigogine’s sense. In these dynamical processes one might expect
that at certain points in time, the trajectory of the system in state space (the
evolving scenario), bifurcates into multiple asymptotically stable (attractors) or
unstable (repellers) regimes in ways depending on critical values of the control
parameters, which stand for the coupling among variables (Nicolis, 1991).
Problem solving, as a creative process, proceeding through regressive
steps could also be seen as following a punctuated equilibrium model, where
long periods of stasis or slow changes are alternated by short periods of rapid
changes. These changes could be seen as state transitions and viewed as catas-
trophes controlled by variables, such as M-capacity, logical thinking or field de-
pendence, when the cognitive task demands it. The trajectory possesses a ple-
thora of successive bifurcation points until the process converges. A cascade of
bifurcations has a cumulative effect on the system’s state and it might cause the
mind to precede successfully towards a creative outcome or be misled and trap-
ped into a less optimal attractor (failure). This cumulative effect is what could
finally be captured by the cusp catastrophe stochastic model at the behavioral
model.
Research Hypotheses
There are four interdependent research hypotheses associated with the
present work:
1. The three cognitive variables: information processing capacity, field
dependence- independence and logical thinking, operationalizing the
corresponding constructive operators can be implemented in a nonlinear cups
catastrophe model.
2. The degree of field dependence-independence and logical thinking
act as bifurcation variables in a cusp catastrophe model, where the information
processing capacity acts as the asymmetry factor.
3. The bimodal distribution of students’ achievement could be modeled
and explained by implementing catastrophe theory models.
4. Problem solving is a nonlinear dynamical process.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 155
METHOD
Participants and Measures
The subjects (N =86, 48% male) were students in the 10th grade high
school, aged 16, who were taking (among others) a compulsory course in phy-
sics. The sample consisted of students from two different schools located in the
broad area of Athens. Data were collected during one school year through paper-
pencil tests. Students’ achievement scores used as the dependent variable were
from a problem-solving task, which was part of their final examination.
Test 1 (Time 1) included knowledge recall questions from related
theory, simple calculations (partial steps) and simple conceptual questions. The
above assured that a minimum of basic prerequisite knowledge was available.
Since students’ failure might be due to lack of information or recall, it is a pre-
requisite that the subjects are familiar with specific domain knowledge. Test 2
(Time 2) was a demanding problem in Newtonian physics. Both Test 1 and Test
2 were parts of the same examination paper. Thus, time is implicit here. The
working hypothesis behind the rationale is whether one could predict students’
performance in problem solving by knowing their achievements in the
prerequisite theoretical knowledge and their psychometric scores.
In addition, all students were assessed for the following three psycho-
metric variables: (a) M-capacity was assessed by means of the Figural
Intersection Test (Pascual-Leone & Burtis, 1974). The test was used for
measuring the functional M-capacity of the subjects according to the method of
Niaz (Pascual-Leone & Burtis, 1974; Niaz 1988a). Cronbach’s alpha reliability
coefficient for the present sample was 0.88. The range of students’ scores was
2.5 - 7.0 (min 0, max 9) and a mean functional M-capacity of 5.10 (SD = 1.22)
was calculated. (b) Cognitive style - Disembedding ability: Disembedding ability
is usually assessed by means of the Group Embedded Figures Test, GEFT,
however, a similar test, the ‘Hidden-Figures Test’ (HFT) was used, which has
been devised and calibrated by Johnstone’s group (1986, 1989) from Witkin’s
original test materials. This test had a scale from zero to 18. The range of
students’ scores was 2.0 – 15.0 (min 0, max 18) and the mean was calculated
9.87 (SD = 2.87). Cronbach’s alpha reliability coefficient was 86. (c) Logical
Thinking (or developmental level): It was assessed by the Lawson test of formal
reasoning (Lawson, 1978). The range of students’ scores was 10 – 100 (min 0,
max 100) and the mean was calculated 59.3 (SD = 28.9). A Cronbach’s alpha
reliability coefficient of 0.83 was obtained.
In order to accommodate Fig. 5 subjects were categorized in four
grades: Field Independent (all > 15), Field medium low (from 15 to 12), Field
medium high (from 11 to 7) and Field Dependent (all < 7). In this test, the lower
values correspond to field dependent subjects and the higher values correspond
to field independent subjects. Note that in the present cusp catastrophe model,
one has to reserve high values for field dependent subjects. Thus, one could
either reverse the scale or make a proper correction for location by assigning the
156 NDPLS, 15(2), Stamovlasis
zero at the maximum value. An analogous categorization was made in order to
accommodate Fig. 6.
Statistical Analysis
In the present analysis the nonlinear regression method was implement-
ed (Guastello, 1987, 1995, 2002). The dependent measure in this study was the
change in students’ achievement from Time 1 to Time 2 (Time is implicit here).
Students’ raw scores were transformed to Z scores corrected for location and
scale σs:

Z = (Y – Ymin) / s (1)
Location correction is made by setting the zero at the minimum value of Y and
the scale was measured by the ordinary standard deviation, s.
The specific equation to be tested for a cusp catastrophe model is:
Δ Z = Z2 - Z1 = b1 Z13 + b2 Z 12 + b3 F Z 1 + b4 M + b5 (2)
Δ Z = Z2 - Z1 = b1 Z13 + b2 Z 12 + b3 L Z 1 + b4 M + b5 (3)

Δ Z = Z2 - Z1 = b1 Z13 + b2 Z 12 + b3F F Z 1 + b3L L Z 1+ b4 M + b5 (4)

The alternative linear models are the following:
Linear 1: ΔZ = b1 M + b2 B + b3 (5)

Linear 2: ΔZ = b1 M + b2 B + b3 M B + b4 (6)
Linear 3: Z2 = b1 M + b2 B + b3 Z1 + b4 (7)
Z is the normalized behavioural variable, while M and B are the normalized
asymmetry and bifurcation respectively, with B= F or L. The variables entered
simultaneously in the Least Squares (LS) procedure.
RESULTS AND INTERPRETATION
Figure 4 shows frequency distribution of scores in Test 1 (Time 1) and
Test 2 (Time 2). Test 1 comprises a mental task of low demand and gives a
unimodal distribution with relatively small variance (mean = 6.01 and SD =
0.95). Test 2 comprises a task of high mental demand and gives a bimodal
distribution with larger variance (mean = 3.91 and SD = 1.85).
In Fig. 5 and Fig. 6, which shows the graphical representation of
achievement frequencies versus the different levels of field dependence-
independence and logical thinking (bifurcation variable) respectively, the
bimodality of the system as a weak ‘catastrophe flag’ can be noticed. However,
a ‘catastrophe flag’ provides merely an indication and it is not sufficient for
demonstrating a catastrophe effect.
 ND
DPLS, 15(2), Nonlinearity
N an
nd Self-organiization in Probblem Solving 1157

Fig. 4. Frequency dis

stribution of stu
udents’ achieve
ement at Time 1 and Time 2.

Fig. 5.. Achievementt with differentt levels of Fie

eld Dependencce-Independen
nce
(bifurca
ation variable).
158 NDPLS, 15(2), Stamovvlasis

Fig. 6. Achievement with

w different le
evels of Logicall Thinking (bifu
urcation variable)
In order to test
t the nonlineear hypothesiss, that a cusp ccatastrophe is tthe
appropriate model to o describe studdents’ achievem ments the regression Eqs. 2, 3,
and 4 should
s accountt for a larger percent
p of the vvariance in thee dependent vaari-
able, thhan the linear counterparts Eqs.
E 5, 6 and 77. In addition,, in order for tthe
cusp too be the appro opriate model for the data bboth the cubic and the produuct
terms ini Eqs. 2-4 sho ould have signnificant weightts and the connfidence intervals
(95% CI)
C should not span the zero point.
p
Table 1 show ws the regressiion slopes, stanndard errors, t--tests, confidennce
intervaals and model fitf for cusp cataastrophe modeel and the contrrol linear models
for men ntal capacity (MM) and field deependence-indeependence (F).
The first lineear model baseed on Eq. 5 acccounted for alm most none of tthe
variancce in the depen ndent variable and the weighht of F was innsignificant, evven
thoughh the weight off M-capacity was w statisticallyy significant (F F(2, 83) = 2.8, p
<0.05).. The second liinear model baased on Eq. 6 aalso accountedd for almost noone
of the variance
v in thee dependent varriable and the weight of the iinteraction effe fect
M X F was insignificcant, even thou ugh F was statiistically signifficant (F(3, 82)) =
2.68 p <0.05).
The third co ontrol model based
b on Eq. 77, was able to ppredict a signiifi-
cant poortion of the vaariance in the dependent
d variiable (Adjustedd R2 = 0.29, F F(3,
82) = 21.5,
2 p <0.000 01). The weigh hts of Z1 and F were insigniificant, while tthe
weight of M-capacity y was (t(82) = 5.65,
5 p <0.00001) and thus, thhey can be connsi-
dered as
a predictors within
w the linearr model explainning 29% of thhe variance in tthe
depend dent variable.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 159
Table 1. Regression Slopes, Standard Errors, t-tests, Confidence intervals and
Model Fit for Cusp Catastrophe Model and the Controls Linear Models. M-
capacity (M) and Field-Dependence (F).

Adj
Model R2 b seb t 95% CI Model F
Linear 1 .03 2,8*
M 0.017 0.008 2.02* 0.008 0.025
F -0.002 0.009 -0.22 -0.011 0.008

Linear 2 .04 2.68*

M 0.291 0.220 1.32 0.007 0.050
F 0.433 0.186 2.32* 0.023 0.058
MXF -0.327 0.212 -1.54 -0.002 -0.000

Linear 3 .29 21.5****

Z1 0.119 0.077 1.54 0.041 0.196
M 0.036 0.006 5.65**** 0.029 0.042
F -0.009 0.006 -1.4 -0.015 -0.003

Cusp .49 37.1****

Z13 0.134 0.032 4.20**** 0.102 0.166
Z12 -0.783 0.150 -5.21**** -0.934 -0.633
F X Z1 -0.002 0.002 -1.05 -0.003 -0.000
M 0.036 0.006 5.68**** 0.029 0.042
* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001,
The cusp model, based on Eq. 2 was able to predict a significant por-
tion of the variance in the dependent variable, change in achievement from Time
1 to Time 2 (Adjusted R2 = 0.49, F(4, 81) = 37.1, p < 0.0001). Three of the terms
in the basic cusp model significantly predicted change in the behavioral vari-
able: the cubic term (t(81) = 4.20, p < 0.0001), the quadratic term (t(81) = -5.21,
p < 0.0001), the asymmetry parameter M-capacity (t(81) = 5.68, p < 0.0001),
while the bifurcation parameter, F was insignificant. Table 2 shows the regres-
sion slopes, standard errors, t-tests, confidence intervals and model fit for cusp
catastrophe model and the controls linear models for mental capacity (M) and
logical thinking (L).
160 NDPLS, 15(2), Stamovlasis
Table 2. Regression Slopes, Standard Errors, t-tests, Confidence intervals and
Model Fit for Cusp Catastrophe Model and the Controls Linear Models. Mental
capacity (M) and Logical Thinking (L).

Model Adj R2 b seb t 95% CI Model F

Linear 1 .12
M -0.003 0.009 -0.36 -0.011 0.006 11.4***
L -0.050 0.012 -4.08*** -0.062 -0.038

Linear 2
M .12 0.032 0.014 2.29* 0.018 0.045 11.4***
L 0.027 0.027 1.00 0.000 0.054
MXL -0.003 0.001 -3.17** -0.003 -0.002

Linear 3 .43
Z1 0.112 0.069 1.64 0.043 0.180 40.1****
M 0.017 0.0062 2.75** 0.010 0.023
L -0.054 0.008 -6.43**** -0.062 -0.046

Cusp .60
3
Z1 0.112 0.028 3.94**** 0.012 0.024 56.3****
2
Z1 -0.649 0.134 -4.86**** 0.083 0.140
L X Z1 -0.012 0.002 -6.32**** -0.783 -0.515
M 0.018 0.006 3.02** -0.014 -0.010
* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.
The first linear model based on Eq. 5 accounted for 12% of the vari-
ance in the dependent variable and the weight of M was insignificant, even
though the weight of L was statistically significant (F(2, 83) = 11.4, p < 0.001).
The second linear model based on Eq. 6 also accounted for 12% of the variance
in the dependent variable and the weight of M the interaction effect M X L were
significant, even though L was statistically insignificant (F(3, 82) = 11.4 p <
0.001).
The third control model based on Eq. 7, was able to predict a
significant portion of the variance in the dependent variable (Adjusted R2 = 0.43,
F(3, 82) = 40.1, p < 0.0001). The weights of Z1 was insignificant, while the
weights of M-capacity (t(82) = 2.75, p < 0.01) and L (t(82) = -6.43, p < 0.0001)
 ND
DPLS, 15(2), Nonlinearity
N an
nd Self-organiization in Probblem Solving 1161
were in nsignificant, and
a thus, they can be considdered as prediictors within tthe
linear model
m explaining 43% of the variance in thee dependent vaariable.
The cusp mo odel, based on n Eq. 3 was abble to predict a significant poor-
tion off the variance in the dependeent variable, a change in acchievement, froom
Time 1 to Time 2 (Ad djusted R2 = 0.60, F(4, 81) = 56.3, p < 0.00001). All termss in
the bassic cusp modell significantly predicted channge in the behhavioral variabble:
the cubbic term (t(81) = 3.94, p < 0..0001), the quaadratic term (t((81) = -4.86, p <
0.0001), the asymmeetry parameter, M, (t(81) = 3 ,02, p < 0.01) and the bifurcca-
tion parameter, L, (t(881) = -6.32, p < 0.0001).
The cusp model,
m based on n Eq. 4 with booth bifurcationn variables, F aand
L present did not im mprove the model,
m thus it is not considdered as a bettter
candidaate and it is noot presented. Finally,
F the prooposed model is the one bassed
on Eq.. 3, with logiccal thinking, L, L as bifurcatiion and its thhree dimensionnal
represeentation is depiicted in Fig.1.
Moreover, cusp
c catastroph he models weere tested for cases where tthe
Time 2 mental task was simple ap pplication andd algorithmic pproblem. In booth
cases thhe cusp modell was not supeerior to the lineear counterpartt. The results aare
summaarized in Fig. 7,7 which showss the adjusted R 2 versus the ccomplexity of tthe
problemm for immediaate comparison with the demaanding problem m case.

odel adjusted R2 versus comp

Fig. 7. Change of mo plexity of the prroblem. The cu
usp
ophe becomes
catastro s superior to pre
e-post linear m
model in deman nding problemss.
DIISCUSSION: THEORETIC
T CAL PERSPE CTIVES ON PROBLEM
SOL
LVING AND CREATIVE
C P
PROCESSES
Conclusio
ons on Hypoth
heses
Generally, we
w can conclud de on the initiaal research hyppotheses that: (a)
The ovverload of the limited
l human channel capaccity could be uunderstood bettter
by refeerence to disem
mbedding abilitty and to logiccal thinking. (bb) Bimodal disttri-
162 NDPLS, 15(2), Stamovlasis
bution of students’ achievement could be explained by implementing a cusp
catastrophe model. (c) In the present data, the information processing capacity is
the asymmetry factor and logical thinking acts as a bifurcation parameter in a
cusp catastrophe model, while for field dependence-independence the hy-
pothesis was not supported by the data, as initially it had been hypothesized.
The results of the present work add to the research endeavor, which in-
vestigates nonlinear effects of neo-Piagetian constructs in problem solving. The
nonlinear interactions of variables in the model suggest that the operationalized
mental recourses could interact in task executions with each other in a nonlinear
fashion, and this could lead to abrupt and unexpected changes from success to
failure. Since the dependent measure is the change in achievement, it should be
emphasized that discontinuities are most likely to appear when the two tasks, at
Time1 and Time2 have a significant demand difference. As shown in Fig. 7, the
nonlinear model becomes statistically significant when the problem is demand-
ing. Abrupt changes, catastrophes, are not expected to be observed if the
demands of the two tasks are comparable. This could be explained if the chal-
lenging task (Time 2) viewed as a ‘random force’ and an ‘entropy-inducing
event’, causing dispersion of mental resource on the mark scale and the variance
of the resulting score distribution to increase. From the NDS perspective, in such
cases a bifurcation mechanism has the chance to operate and reduce the entropy
of the system by partitioning the energized elements into attractor regions of rel-
ative stability or equilibrium (Guastello, 2002; Thompson, 1982).
The cusp catastrophe model introduces nonlinearity to neo-Piagetian
theories based on empirical evidences. In the present research, the hypotheses 1-
3 are directly tested from the statistical analysis. The answer, however, to hypo-
thesis 4, that is ‘problem solving is a nonlinear process’ is based not merely on
hypotheses 1-3, but it is supported by the following theoretical analysis, which
emphasizes the dynamical hypothesis, not as a metaphor, but as a causal inter-
pretation of observed bifurcation effects.
Co-evolution of Methodology, Epistemology and Theory
Traditionally in the philosophy of social science, the methodological
approaches are connected through linear ‘causal’ relationships with some basic a
priori philosophical positions. That is, starting arbitrarily from some ontological
pre-assumptions about the human social reality, one may formulate the relative
epistemological consideration and then make the appropriate methodological
choices1, which further determine the theory development (if any). However,
what has been shown with the emerged new paradigm of complexity2, methodo-
logical advancements have affected ontological and epistemological assump-
tions, that is, a circular causality or a co-evolution of ontology, epistemology
and methodology has been demonstrated (Allen, 2006).
The development of nonlinear statistical methods has provided the pos-
sibility of inductively detecting nonlinear effects at the behavioral level. Con-
cerning the present research, whereas the linear models can only provide em-
pirical evidence for the role of neo-Piagetian operators supporting the construct
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 163
validity of the TCO, the nonlinear statistical approach adds more to our
knowledge. It reveals bifurcation and hysteresis effects in a cusp catastrophe
structure and suggests a pattern of interactions.
The cusp model in the present application reveals the pattern of be-
havior (achievement) in problem solving, showing the functional dependence of
the chosen control parameters, M and L. The model demonstrates that both lin-
ear and nonlinear changes in behavioral variable might be expected and how
these changes could be described by the control variables, L and M: At high val-
ues of L changes are smooth and at low values of L they are discontinuous. At
low values of M changes occur over the lower mode and are relatively small. At
middle values of M, changes occur between modes and are relatively large. At
high values of M, changes occur around the upper mode and are again small. At
the control surface we can observe the bifurcation set mapping in the unfolding
of the surface in two dimensions. The cusp bifurcation set induces two diverging
response gradients, which are joined at a point, the cusp point. At the cusp point
the behavior is ambiguous, while the two diverging gradients represent a varying
degree of probability that a student might succeed or fail (Figs. 1 and 2). The
phenomenon of hysteresis observed, that is, subjects with the same parameter
values (M, L) oscillating between stable states, means that small differences in
M or L may lead to sudden jumps from success to failure (Fig. 3).
The above geometry of behavior demonstrates the nonlinear interaction
between variables operationalizing neo-Piagetian constructs. The nonlinearity
observed at the behavioral level is phenomenological, which apparently
describes rather than explains the behaviour. However, as it will be shown by
the following epistemological discussion, a nonlinear dynamical model implies
causal explanation.
Moreover, a nonlinear phenomenology, which is revealed due to
methodological advancements, raises a series of philosophical questions. First, it
raises ontological questions about the nature of mental processes involved in
problem solving. Second, it questions the epistemology of the traditional linear
science. It is important to repeat here that the neo-Piagetian theories and the
information processing models explicitly or implicitly assume the computer
metaphor for mind and this justifies the implementation of linear models. How-
ever, the linear, cause-and-effect paradigm does not always hold when applied
on factors affecting human performance. Knowledge acquisition, problem solv-
ing and creativity are more complex processes when considering the variety of
ability levels, patterns of thinking, learning styles, personality traits, and cultural
backgrounds. The composite profile of humans and their achievement outcomes
cannot be simple weighted linear sums of the contributed components. These
components are rather interacting in task executions, with each other in a
nonlinear fashion, and this could lead to surprises and different qualitative out-
comes, especially when a creative process is considered. Thus, under the new
perspective, the nonlinear paradigm, the action of constructive operators
representing mental sub-processes, should be adopted within a different scenario
in order a nonlinear phenomenology to be explained.
164 NDPLS, 15(2), Stamovlasis
Therefore, what is needed is, first, an understanding through a verbal
description, of the likely nonlinear characteristics of mental processes (which
has already been presented in the introduction section) and second a formal
description, that is, a description in a mathematical language, of a causal
interpretation of the discontinuities observed in problem solving. This will
explain the phenomenology of bifurcation and it will shed light on the nature of
these cognitive processes. A mathematical foundation of such a causal
interpretation is central for the ‘hard core’ elements of the new science (Lakatos,
1974). To fulfill the above claim, an attempt is made next by employing the
mathematical theory of self-organization, which advances the theoretical
perspective of the neo-Piagetian framework in creative mental processes and
merges it with nonlinear dynamics and complexity.
A Causal Interpretation for the Nonlinear Phenomenology
of a Creative Process
The main characteristic of a cohesive scientific theory is its mathemati-
cal foundation. NDS and complexity have been used as sources of metaphors for
describing phenomena evolving in time, but beyond this description, mainly in
physical and biological science, they have served as a general and rigorous
framework with explicit mathematical tools for testing hypotheses. The founda-
tion of NDS and complexity in cognitive phenomena, such as, problem solving
and creativity, also needs to be mathematically supported. The main scientific
theories of nonlinear dynamics: Prigogine’s non-equilibrium thermodynamics
(G. Nicolis, & Prigogine, 1977; Prigogine, 1961), Hakens’s synergetics (1983,
1990) and Thom’s catastrophe theory (1975), even though they were grown with
different rationales, they have fostered the central notion of self-organization
and focused on its mathematical development. The important feature of complex
systems is the emergent properties through self-organization processes and the
phenomenology of bifurcations, which correspond to phase or state transitions
for physical or human systems respectively. Bifurcations are signatures of com-
plexity and indicative of self-organization mechanisms (Nicolis & Nicolis,
2007).
In mental phenomena under investigation, creative processes of
problem solving, the main concern is the emergence (creation) of outcomes: a
new pattern, new information or a solution. The emergence of such an outcome
is a discontinuity, thus a state transition for the cognitive system. Self-
organization can serve as theoretical interpretation of bifurcations and state
transitions within a causal nonlinear dynamical perspective. Self-organization
has been successfully implemented for physical and biological systems and it
could be transferred to human system as well, for explaining emergent patterns
observed in psychological processes (Hollis, Kloos & van Orden, 2008). It has
been fostered for a causal interpretation of Piaget’s theory of stepwise cognitive
development (Molenaar & Raijmakers, 2000).
Let’s consider brain operating in a mental task execution and function-
ing as an n-dimensional system evolving in time. Then, at any point in time the
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 165
state of the system could be described by an n-dimensional vector and its
geometrical representation in the state space. As a property of complex systems,
there are special points or regions in the state space, where the rate change of the
system is zero and the system appears to be stable to external perturbations.
These are the equilibrium states of the system known as the system’s attractors.
Suppose a problem solving process, that is time-dependent changes of
the state of the n-dimensional system under investigation, and it is described by
n-coupled mathematical (however unknown) equations, the evolution equations.
These express the rate of change of the state at a given time t as a function of the
state at time t-1, with the function depending on a set of parameters. The
equilibria set of the evolving process, the system’s attractors, also depend on the
parameters because in a mathematical sense they are actually the solutions of the
evolution equations. By definition in the equilibrium states the rate change of
the system is zero. Thus the topology of the attractors in a dynamical system
depends, among others, as well as on the parameters of the system.
Let parameters change smoothly with time. The question is, what the
impact of these changes to the topology of the equilibrium set, is. From a
geometrical point of view smooth parameter changes in a linear evolution
equation result in a small change of the equilibrium set. However, if the
evolution equation is nonlinear, small parameter variation could give rise to
sudden changes in the topology of the equilibrium set. Small changes of the
parameter could give rise to qualitative changes in the topological pattern. Thus,
existing equilibria might disappear in favor to new ones that emerge as a result
of the small changes in the parameters. That is mathematically a bifurcation and
a state transition for the psychological system.
The parameter variations however, are necessary but not sufficient for
the bifurcations to occur and to determine the qualitative characteristics of the
new attractors’ topology. Bifurcation mechanism in a physical system such as
Zeeman’s catastrophe machine is nested in the operation of a potential function
and the dynamics of the system (Zeeman, 1976). Similarly, not only the change
of the parameters, but the dynamic behavior of the cognitive system is the for-
mative cause of the ensuing bifurcation and the emergence of the new topologi-
cal pattern in the state space of the system. The nonlinear dynamic behaviour
itself gives the system the capacity for self-organizing transformation. The men-
tal process under study, a creative process, constitutes a self-organization pro-
cess, where through bifurcations, new topological pattern appears, that is new
information is produced. The mathematical theory of self-organization matches
the characteristics of a metal creative process in problem solving, where a new
idea, a solution or an innovative outcome emerges.
A reasonable question, however, is how one may analyze
mathematically such mental process, which is expected to be non-ergodic, not
experimentally-tractable, and moreover, if the underlying evolution equation is
unknown. The good news, however, is that for a certain class of bifurcations, the
elementary catastrophes, as catastrophe theory exemplifies (Castrigiano &
Hayes, 2004; Gilmore, 1981; Poston & Stewart, 1978; Thom, 1985), the
166 NDPLS, 15(2), Stamovlasis
evolution equation does not have to be known and the analyses of the local
observed behaviour can be carried out with a small number of control
parameters. For a psychological system where numerous variables could be of
importance, the option that large changes or transitions could be locally
controlled and described by only a few variables is certainly an advantage. This
point singles out the applicability of CT in psychology and in particular to men-
tal processes, which can exhibit emergent phenomena.
The new epistemology rising from the bridges developed between neo-
Piagetian theory and NDS and founded on mathematical theory of self-organiza-
tion might have to face ontological questions about the origin of nonlinearity
evident at the behavioral level. Appropriate answers could be easily supported
by the advances of contemporary neuroscience, which built on the dynamical
systems approach and on the science of complexity (Freeman, 2000; Freeman &
Barrie, 2001) and especially on theoretical models, which consider the brain
functioning as a dissipative dynamical system possessing multiple attractors
(Nicolis & Tsuda, 1999).
Moreover, epistemological questions about the compatibility of the
potential function, which is assumed in the mathematical formalization of a
catastrophe model, can be answered similarly. The mathematical theory of self-
organization for a dynamical system possessing multiple attractors, suggests that
in the vicinity of bifurcations, and given that the evolution equation is unknown,
a potential function with multiple stable equilibria is not only reconcilable, but it
is the proper modeling tool.
Conclusions, Implications and Directions for Future Research
The present research endeavor builds bridges between NDS and crea-
tive processes, such as problem solving, and adds to the framework of catas-
trophe theory applications with neo-Piagetian constructs. Methodologically, the
work started inductively, by analyzing empirical data and demonstrating com-
plex-systems phenomenology (Nicolis & Nicolis, 2007) and then, NDS was in-
vited and presented in a theoretical analysis, as its causal interpretation. More-
over, a formal description of creative mental processes was presented by imple-
menting the mathematical theory of self-organization. The present work adds to
the paradigm shift realized in social science and particularly in cognitive and
developmental theories (Lewis, 1994, 2000; Molenaar & Raijmakers, 2000;
Thelen & Smith, 1994; van der Maas & Hopkins, 1998; van Geert, 1991, 2003).
The large body of research for psychological science consolidated with NDS
and Complexity has already manifested a cohesive theoretical framework for
further deductive endeavors.
The amalgamation of NDS with neo-Piagetian theories has important
implications for the latter and educational sciences as well. A fundamental as-
pect in the new perspective is the view change about phenomena under concern,
such as learning and creativity, based on a new philosophical –ontological and
epistemological- shift. The learning stage of a mind (as a dynamical system) is
characterized by an attractor, which is a behavioral mode (performance) as a
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 167
function of interacting internal components of the system and reflects its
potential. Problem solving, innovation and creativity imply the emergence of a
new attractor that is a state transition realized via a nonlinear dynamic process
and self-organization mechanisms. Such view affects our understanding of how
creative processes work and transforms the definition and the essence of learn-
ing with important educational implications (see Stamovlasis, 2006).
Future research should be directed towards extending the current find-
ings to a variety of circumstances. There are a lot of psychometric variables
from neo-Piagetian framework, and not merely, such as working memory capa-
city, convergent-divergent thinking, mobility fixity dimension or other cognitive
styles and personality traits, which are known as predictors of creative behavior
in a range of disciplines. Identification of potential bifurcation variables for a
variety of mental tasks will be a crucial step towards understanding inhibitory
processes of creativity and will shed light on mechanisms of emerging
phenomena.
Last, but not least, besides the promising merger, the Nonlinear Dyna-
mical Systems theory for the mental processes under investigation attains addi-
tional support by a priori arguments. These arguments originate from the neces-
sity to fill the gap between nature versus nurture or genetic versus environment
dilemmas. The neo-Piagetian constructs in TCO, which represent mental re-
sources, have neural origin (Grigsby & Osuch, 2007) and thus are related to
brain functioning and structure. This implies a partial effect of genetic factors.
However, genetic theories, as in the case of human development (Molenaar &
Raijmakers, 2000) cannot fully account for high ability and creativity. The infor-
mation for ‘ability’ in problem solving or creativity cannot be stored in the
genome, because it would require enormous capacity. On the other hand, envi-
ronmental factors alone have not shown to be always effective in ameliorating
individual differences.
The latter still remains a great challenge for education (environment),
that is, how to take advantage of the knowledge sourcing out of the new para-
digm and apply it to educational practices. The dynamical hypothesis has come
into view in theoretical approaches related to education and pedagogy
(Abraham, 2003; Finke, & Bettle, 1996; Fromberg, 2010) and at a metaphorical
level the NDS has already influenced new proposals for postmodern curricula
and educational practices (Doll, 1993; Fleener, 2002; Davis, Smith, & Leflore,
2008). However, it is important that research in educational sciences applies
concepts and techniques of NDS to specific research problems and builds upon
the new paradigm of complexity.
Education and educational research might be the crucial fields, where
the development of the new paradigm is prerequisite for realization of the global
paradigm shift in society, that is, the total world-view change, determined by
shifts in ethical and moral grounds (Fleener & Merritt, 2007).
ENDNOTES
1
It is written in the most introductory Methodology textbooks. It refers
168 NDPLS, 15(2), Stamovlasis
to two antithetic philosophical assumptions about the nature of society and
social science: realism-positivism-determinism versus nominalism-anti-
positivism-voluntarism and to subsequent methodological choices, Idiographic
versus Nomothetic approaches.
2
The Complexity is referred as a new paradigm for physical and
biological sciences (Nicolis, & Nicolis, 2007), and in some areas of social
sciences, based on shifts made in epistemological and ontological
considerations.
ACKNOWLEDGEMENT
The author is in debt to Dr. Andreas Dimos for his significant
contribution in the evaluation of psychometric tests. In addition, the author is
also thankful to Maria Birbili for her constructive comments on the manuscript,
to the Science Teachers of ‘Λύκειο Πόρου’, C. Modiatis and Μ. Brokos for
permitting the data collection in their classes and especially to C. Sarpa, the
School Principal of ‘Εκπαιδευτήρια Ζώη’, for her invaluable cooperation.
REFERENCES
Abraham, J. (2003). Dynamical system theory: Application to Pedagogy. In W.
Tschacher, & J. P. Dauwalder (Eds.), The dynamical systems approach to
cognition (pp 295-307). Singapore: World Scientific.
Allen, P. (2007). The co-evolution of epistemology, axiology and ontology. Nonlinear
Dynamics, Psychology and Life Science, 11, 19-50.
Baddeley, A. D. (1986). Working memory. Oxford: Oxford University Press
Baker, J. S., & Frey, P. W. (1980). A cusp catastrophe: Hysteresis, bimodality, and inac-
cessibility in rabbit eyelid conditioning. Learning and Motivation, 10, 520-535.
Bernardez, E. (1995). On the study of language with tools of catastrophe theory. Atlantis,
XVII, 1-2, 261-291.
Boles, S. (1990). A model of routine and creative problem solving. Journal of Creative
Behavior, 24, 230-245.
Case, R. (1985). Intellectual development: Birth to adulthood. San Diego: Academic
Press.
Case, R. (1978). Piaget and beyond: Toward a developmentally based theory and
technology of instruction. In R. Glaser (Ed.), Advances in instructional
psychology (pp. 167–228). Hillsdale: Lawrence Erlbaum.
Castrigiano, D. P. L., & Hayes, S. A. (2004). Catastrophe theory. Boulder: Addison-
Wesley Publishing.
Cobb, L., Koppstein, P., & Chen, N. (1983). Estimation and moment recursion relations
for multimodal distributions of the exponential family. Journal of the American
Statistical Association, 78, 124–130.
Cobb, L., & Watson, B. (1980). Statistical catastrophe theory: An overview.
Mathematical Modelling, 1, 311–317.
Cobb, L., & Zacks, S. (1985). Applications of catastrophe theory for statistical modeling
in the biosciences. Journal of the American Statistical Association, 80, 793-
802.
Davis, E. J., Smith, T. J., & Leflore, D. (2008). Chaos in the classroom: A new theory of
teaching and learning. Durham: Carolina Academic Press.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 169
Danili, E., & Reid, N. (2006). Cognitive factors that can potentially affect pupils’ test
performance. Chemistry Education Research and Practice, 7, 64-83.
Doll, E. W. (1993). A post modern perspective on curriculum. New York: Teachers
College Press.
Edelman, G. M. (1987). Neural Darwinism: The theory of neuronal group selection. NY:
Basic Books.
Edelman, G. M. (1989). The remembered present: A biological theory of consciousness.
NY: Basic Books.
Feraro, T. (1978). An introduction to catastrophes. Behavioral Sciences, 23, 291-316.
Finke, R. & Bettle, J. (1996). Chaotic cognition. New Jersey: Lawrence Erlbaum
Associates.
Fleener, J. (2002). Curriculum dynamics: Recreating heart. New York: Peter Lang.
Fleener, J., & Merritt, M. (2007). Paradigms lost? Nonlinear Dynamics, Psychology and
Life Science, 11, 1-18.
Freeman, W., & Barrie, J. M. (2001). Chaotic oscillations and the genesis of meaning in
cerebral cortex.. In W. Sulis, & I. Trofimova (Eds.), Nonlinear Dynamics in the
Life and Social Science (pp. 45-62). NATO Science Series, Vol. 320.
Freeman, W. (2000). How brains make up their minds. NY: University Columbia Press.
Fromberg, D. P. (2010). How nonlinear systems inform meaning and early education.
Nonlinear Dynamics, Psychology and Life Science, 14, 47-68.
Castrigiano, D. P. L. &. Hayes, S. A. (2004). Catastrophe theory. Boulder: Addison-
Wesley Publishing.
Gilmore, R. (1981). Catastrophe theory for science and engineers. New York: Wiley.
Grasman, R., van der Maas, H. & Wagenmakers, E. (2009). Fitting the cusp catastrophe
in R: A cusp package primer. Journal of Statistical Software, 32, 1-27.
Grigsby, J. & Osuch, E. (2007). Neurodynamics, state, agency, and psychological
functioning. In C. Piers, J. P. Muller, & J. Brent (Eds.), Self-organizing
complexity in psychological systems (pp. 37-83). New York: Jason Aronson.
Guastello, S. J. (1982). Moderator regression and the cusp catastrophe: Application of
two-stage personnel selection, training, therapy, and program evaluation.
Behavioral Science, 27, 259-272.
Guastello, S. J. (1987). A butterfly catastrophe model of motivation in organization:
academic performance. Journal of Applied Psychology, 72, 165-182.
Guastello, S. J. (1995). Chaos, catastrophe, and human affairs: Application of nonlinear
dynamics to work, organizations, and social evolution. Mahwah, NJ: Lawrence
Erlbaum Associates.
Guastello, S. J. (2001). Nonlinear dynamics in psychology. Discrete Dynamics in Nature
and Society, 6, 11-19.
Guastello, S. J. (2002). Managing emergent phenomena: Nonlinear dynamics in work
organizations. Mahwah, NJ: Lawrence Erlbaum Associates.
Haken, H. (1983). Synergetics, nonequilibrium phase transition and self-organization in
physics, chemistry, and biology. Berlin: Springer-Verlag.
Haken, H. (1990). Information and self-organization: A microscopic approach to
complex systems. Berlin: Springer-Verlag.
Hollis, G., Kloos, H., & van Orden, G. (2008). Origins of order in cognitive activity. In S.
Guastello, M. Koopmans, & D. Pincus (Eds), Chaos and complexity in
psychology: The theory of nonlinear dynamical systems (pp. 206-241). NY:
Cambridge University Press.
170 NDPLS, 15(2), Stamovlasis
Johnstone, A. H., & El-Banna, H. (1986). Capacities, demands and processes - A
predictive model for science education. Education in Chemistry, 23, 80-84.
Johnstone, A.H., & Al-Naeme, F. F. (1991). Room for scientific thought? International
Journal of Science Education, 13, 187-192.
Johnstone, A.H., Hogg, W.R., & Ziane, M. (1993). A working memory model applied to
physics problem solving. International Journal of Science Education, 15, 663-
672.
Johnstone, A. H. & Selepeg, D. (2000). A Language Problem Revisited. Chemistry
Education: Research and Practice, 2, 19-20.
Kelso, J. A. S. (1995). Dynamic patterns: The self-organization of brain and behaviour.
Cambridge: MIT Press.
Lakatos, I. (1974). Falsification and the methodology of scientific research programmes.
In. L. Lakatos & A. Musgrave (Eds.), Criticism and the Growth of Knowledge
(pp. 100-101). Cambridge: Cambridge University Press.
Lawson, A.E. (1978). Development and validation of the classroom test of formal
reasoning. Journal of Research in Science Teaching, 15, 11-24.
Lawson, A., E. (1982). Formal reasoning, achievement, and intelligence: an issue of
importance. Science Education, 66, 77-83.
Lawson, A., E. (1983). Predicting science achievement: The role of developmental level,
disembedding ability, mental capacity, prior knowledge, and beliefs. Journal of
Research in Science Teaching, 20, 117–129.
Lawson, A.E. (1985). A review of research on formal reasoning and science instruction.
Journal of Research in Science Teaching, 22, 569-617.
Lawson, A.E. (1993). Classroom test of scientific reasoning: Revised paper–pencil
edition. Tempe, AZ: Arizona State University.
Lewis, M. (1994). Reconciling stage and specificity in neo-Piagetian theory: Self-
organizing conceptual structures. Human Development, 37, 143-169.
Lewis, M. (2000). The promise of dynamic systems approach for an integrated account of
human development. Child Development, 71, 36-43.
Miller, A. L. (2007). Creativity and cognitive style: The relationship between field-
Dependence-independence, expected evaluation, and creative performance.
Psychology of Aesthetics, Creativity, and the Arts, 1, 243-246.
Milne1, E. and Szczerbinski, M. (2009). Global and local perceptual style, field-
independence, and central coherence: An attempt at concept validation.
Advances in Cognitive Psychology, 5, 1-26.
Molenaar, P. C., & Oppenheimer, L. (1985). Dynamic models of development and the
mechanistic-organismic controversy. New Ideas in Psychology, 3, 233-242.
Molenaar, P. C., & Raijmakers, M.E. (2000). A causal interpretation of Piaget’s theory of
cognitive development: Reflections on the relationship between epigenesis and
nonlinear dynamic. New Ideas in Psychology, 18, 41-55.
Newell, A. (1995). Unified theories of cognition. Cambridge: Harvard University Press.
Niaz, M. (1988a). The information processing demand of chemistry problems and its
relation to Pascual-Leone's functional M-capacity. International Journal of
Science Education, 10, 231-238.
Niaz, M. (1988b). Manipulation of M-demand of chemistry problems and its effect on
student performance: a neo-Piagetian study. Journal of Research in Science
Teaching, 25, 643-657.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 171
Niaz, M. (1989). Relation between Pascual-Leone's structural and functional M-space
and its effect on problem solving in chemistry. International Journal of Science
Education, 11, 93-99.
Niaz, M. (1995). Progressive transitions from algorithmic to conceptual understanding in
student ability to solve chemistry problems: A Lakatosian interpretation.
Science Education, 79, 19-36.
Niaz, M. (1996). Reasoning strategies of students in solving chemistry problems as a
function of developmental level, functional M-capacity and disembedding
ability. International Journal of Science Education, 18, 525-541.
Niaz, M., & Lawson A. E. (1985). Balancing chemical equations: the role of
developmental level and mental capacity. Journal of Research in Science
Teaching, 22, 41-51.
Niaz, M., & Robinson, W.R. (1992). Manipulation of logical structure of chemistry
problems and its effect on student performance. Journal of Research in Science
Teaching, 29, 211-226.
Niaz, M., de Nunez, S. G., & de Pineda, R.I. (2000). Academic performance of high
school students as a function of mental capacity, cognitive style, mobility-fixity
dimension, and creativity. Journal of Creative Behavior, 34, 18-29.
Nicolis, S. J. (1986). Chaotic Dynamics as applied to information processing. Reports on
Progress in Physics, 49, 1109-1187.
Nicolis, S. J. (1991). Chaos and information processing. Singapore: World Scientific
Publishing.
Nicolis, G, & Prigogine, I. (1977). Self-organization in non-equilibrium systems. NY:
Wiley.
Nicolis, G. & Nicolis, C. (2007). Foundations of Complex Systems. Singapore: World
Scientific.
Nicolis, S. J., & Tsuda, I. (1999). Mathematical describtion of brain dynamics in
perception and action. Journal of Consciousness Studies, 6, 215-228.
Oliva, T., Desarbo, W., Day, D., & Jedidi, K. (1987). GEMCAT: A general multivariate
methodology for estimating catastrophe models. Behavioral Science, 32, 121–
137.
Pascual-Leone, J. (1969). The encoding and decoding of symbols by children: a new
experimental Paradigm and neo-Piagetian model. Journal of Experimental
Child Psychology, 8, 328-355.
Pascual-Leone, J. (1970). A mathematical model for the transition rule in Piaget’s
developmental stages. Acta Psychologica, 32, 301-345.
Pascual-Leone, J. (1987). Organismic processes for neo-Piagetian theories: A dialectal
causal account of cognitive development. International Journal of Psychology,
22, 531–570.
Pascual-Leone, J. (1989). An organismic process model of Witkin’s field dependence-
independence. In T. Globerson & T. Zelniker (Eds.), Cognitive style and
cognitive development (pp. 36-70). Norwood, NJ: Ablex.
Pascual-Leone, J., & Burtis, P. J. (1974). FIT: Figural Intersection Test, a group measure
of M-space. Unpublished manuscript. Ontario: York University.
Pascual-Leone, J. & Goodman, D. (1979). Intelligence and experience. Instructional
Science, 8, 301–367.
172 NDPLS, 15(2), Stamovlasis
Pascual-Leone, J., Goodman, D., Ammon, P. & Subelman, I. (1978). Piagetian theory
and neo-Piagetian analysis as psychological guides in education. In J. M.
Gallagher & J. A. Easley (Eds.), Knowledge and development, (pp. 234–289).
NY: Plenum.
Piaget, J. (1995). The equilibration of cognitive structures. Chicago: The University of
Chicago Press.
Poston, T. & Stewart, I. (1978). Catastrophe theory and its applications. NY: Dover
Publications.
Prigogine, I. (1961). Introduction of thermodynamics of irreversible processes. NY:
Wiley.
Prigogine, I. & Stengers, I. (1984). Order out of chaos: Man's new dialogue with nature.
NY: Bantam.
Preece, P. F. W. (1980). A geometrical model of Piagetian conservation. Psychological
Reports, 46, 142-148.
Saari, D. G. (1977). A qualitative model for the dynamics of cognitive process. Journal
of mathematical Psychology, 15, 145-168.
Scandarmalia, M. (1977). Information processing capacity and the problem of horizontal
decalage: A demonstration using combinatorial reasoning task. Child
Psychology, 48, 301-345.
Stamovlasis, D. (2005). Nonlinearity, thresholds and bifurcations in knowledge
acquisition and problem solving: Towards a paradigm shift in educational
research. In A. Russell, T. Honkela, K. Lagus, & M. Polla (Eds.), Proceedings
of the 1st International and Interdisciplinary Symposium of Adaptive Models of
Knowledge, Language and Cognition (pp. 17-26). Espoo, FL: Otamedia Oy.
Stamovlasis, D. (2006). The nonlinear dynamical hypothesis in science education
problem solving: A catastrophe theory approach. Nonlinear Dynamics,
Psychology and Life Science, 10, 37-70.
Stamovlasis, D. (2008). A catastrophe theory model for the working-memory overload
hypothesis - Methodological issues. In T. Honkela, M. Polla, M. Paukeri, & O.
Simula (Eds.), Proceedings of the 2nd International and Interdisciplinary
Conference on Adaptive Knowledge, Representation and Reasoning (pp. 73-
80). Porvoo, FL: Multiprint Espoo.
Stamovlasis, D. (2010). Methodological and epistemological issues on linear regression
applied to psychometric variables in problem solving: Rethinking variance.
Chemistry Education, Research and Practice, 11, 59-68.
Stamovlasis, D., & Tsaparlis, G. (2003). Nonlinear Analysis of The Effect of Working
Memory Capacity on Student Performance In Problem Solving. In D. Psillos,
P. Kariotoglou, V. Tselfes, E. Hatzikraniotis, G. Fassoulopoulos, & M. Kallery
(Eds.), Science Education Research in the Knowledge Based Society (pp. 183-
190). Netherlands: Kluwer Academic Publishers.
Stamovlasis, D., & Tsaparlis, G. (2005). Cognitive variables in problem solving: A
nonlinear approach. International Journal of Science and Mathematics
Education, 3, 7-32.
Stamovlasis, D., Tsitsipis, G., & Papageorgiou, G. (2010). The effect of logical thinking
and cognitive styles on understanding the structure of matter: An analysis with
the random walk method. Chemistry Education, Research and Practice, 11,
173-181.
Thelen, E., & Smith, L. (1994). A dynamic systems approach to the development of
cognition and action. Cambridge, MA: MIT press.
 NDPLS, 15(2), Nonlinearity and Self-organization in Problem Solving 173
Thom, R. (1975). Structural stability and morphogenesis. NY: Benjamin-Addison-
Wesley.
Thompson, J. M. T. (1982). Instabilities and catastrophes in science and engineering.
NY: Wiley.
Tinajero, C., & Paramo, F. M. (1998). Field dependence-independence cognitive style
and academic achievement: a review of research and theory. European Journal
of Psychology of Education, 13, 227–251.
Tsaparlis, G. (1998). Dimensional analysis and predictive models in problem solving.
International Journal of Science Education, 20, 335-350.
Tsaparlis, G. (2005). Non-algorithmic quantitative problem solving in university physical
chemistry: a correlation study of the role of selective cognitive factors.
Research in Science and Technological Education, 23, 125-148.
Tsaparlis, G. & Angelopoulos, V. (2000). A model of problem solving: Its operation,
validity and usefulness in the case of organic-synthesis problems. Science
Education, 84, 131-153.
Tsitsipis, G., Stamovlasis, D., & Papageorgiou, G. (2010). The effect of three cognitive
variables on students’ understanding of the particulate nature and the changes
of state of matter. International Journal of Science Education, 32, 987-1016.
van der Maas, H. L. J., & Molennar, P. C. (1992). Stagewise cognitive development: An
application of catastrophe theory. Psychological Review, 99, 395-417.
van der Maas, H. L. J., & Molennar, P. C. (1996). The dynamical and statistical
properties of cognitive strategies: Relation between strategies, attractors and
latent classes. In K. M. Newell, & P. C. M. Molenaar (Eds.), Dynamical
systems in development (pp 161-176). Hillsdale, NJ: Erlbaum.
van der Maas, H. L. J., & Hopkins, B. (1998). Developmental transitions: So what's new?
British Journal of Developmental Psychology, 16, 1-13.
van der Maas, H. L. J., Jansen, B. R. J., & Raijmakers, M. E. J. (2004). Developmental
patterns in proportional reasoning. In A. Demetriou & A. Raftopoulos (Eds.),
Cognitive developmental change theories, models and measurement (pp. 118-
156). Cambridge: Cambridge University Press.
van Gelder T. & Port R. (1995). It’s about time: An overview of the dynamical approach
to cognition. In T. Van Gelder & R. Port (Eds.), Mind as motion,exploration in
the dynamics of cognition, (pp 1-43). Cambridge, MA: MIT Press.
van Geert, P. (1991). A dynamic systems model of cognitive and language growth
Psychological Review, 98, 3–53.
van Geert, P. (2003). Dynamic systems approaches and modeling of growth
developmental processes. In J. Valsiner & K. J. Connolly (Eds.), Handbook of
developmental psychology (pp. 640–672). London: Sage.
Witkin, H. A., (1978). Cognitive styles in personal and cultural adaptation. Worcester,
MA: Clark University Press.
Witkin, H. A., & Goodenough, D. R. (1981). Cognitive styles: Essence and origins. NY:
International University Press.
Witkin, H. A., Dyk, R. B., Faterson, H. F., Goodenough, D. R., & Karp, S. A. (1962).
Psychological differentiation. NY: Wiley.
Witkin, H. A., Oltman, P. K., Raskin, E., & Karp, S. A. (1971). A Manual for the
embedded figures test. Palo Alto: Consulting Psychologists Press.
Zeeman, E. C. (1976). Catastrophe theory. Scientific American, 234(4), 65-83.