Dynamic systems approaches are consistent with psychological theory. For example, Mischel and ShodaÕs Cognitive-Affective Processing System of Personality (CAPS) conceptualizes personality as a dynamic system, rather than a set of stable, static, enduring traits that can be seen across situations. Instead, CAPS posits that personality is an open system which produces variability in behavior. The individual and the situation interact to produce behavior, and that behavior subsequently feeds back into the system to influence or change it over time. Additionally, rather than a personÕs behavioral inconsistency reflecting measurement error or low validity of a personality trait, it reflects patterns of variation within a person.
Although the main ideas behind dynamic systems Ð such as in the above example - are common throughout many of the fieldÕs theories, dynamic systems is not a widespread practice within Social Psychology. In a search of studies published in the last 10 years, social psychology articles yield approximately 27,500 articles on Google Scholar. Of these articles, 92 include some reference to dynamic systems in the title. While many early theories in psychology included some element of dynamics within their tenets, current practices and conventions within the field have moved away from these dynamic roots. Indeed, the small rate of dynamic publications within psychology speaks to the ÒgapÓ between how we think about psychological phenomena at an abstract level and how we actually apply dynamic systems techniques to our data at a concrete level.
To better understand the gap, we should first discuss three important historical and logistical factors that may have contributed. First, Dynamic Systems is implicitly built upon the notion that phenomena should be examined as constantly changing. Inherently, there is no other way to understand change then to look at time and space (motion, velocity, acceleration, etc.). However, studying change often requires advances statistical techniques and technology which were not available whilst psychological theories emerged. Similarly, collecting data over time requires repeated measurements. Repeated measures can be either continuous (e.g. measuring cardiovascular impedance) or discrete (e.g. using daily diaries, coding statements in a conversation). Such datasets require greater investigator and participant burden and cost and also typically result in datasets so large and unwieldy, it would be difficult to analyze without existing technology. The advent of computers allowed us to pursue complex research questions with relatively little burden. Thus, the theoretical ideas outpaced the technology available to realistically examine dynamics.
Thus, in order to best meet the ideas of the field in its early days with the tools available, greater experimental control was sought in order to Òisolate effects.Ó Thus, the dominant paradigm, which persists today, is to examine cause and effect relationships between variables in order to control and predict phenomena. This paradigm typically isolates a relationship between variables in a highly controlled experimental/laboratory setting which involves random assignment to study conditions, controlling for confounds and covariates, and by including experimental, control, and comparison study conditions. This approach views phenomena as static rather than dynamic. Now, it appears our technological ability to pursue dynamic system approaches outpaces the dominant paradigm (and thus, dominant training approaches).
|
|
Assumptions
and Approaches |
|
|
Conceptual
Issue |
In
Traditional Psychology |
In
Dynamic Systems |
|
Role of ÒTimeÓ |
Time not included; Usually static |
Time as a critical variable; Dynamic |
|
Role of ÒMovementÓ |
Motion not critical approach to variables [though may be examined via a baseline to task comparison]; Velocity and acceleration not relevant |
Motion, Velocity, and Acceleration of variable key to understanding model |
|
Linearity |
Assumes variables related linearly |
Assumes nonlinearity |
|
What Results? |
Interested in identifying significant associations between variables [either yes/no] |
Interested in identifying patterns, trajectories, or flow in states as revealed by variables of interest [yes/no answer not as crucial] |
|
Crucial Analysis to Examine
Hypothesis/Results |
Null Hypothesis Testing as key (e.g. p-value, confidence intervals) or Model fit statistical tests |
State-space grids; Dynamic topography; Change/time series graphs; Identification of stability/nonstability, periodicity/quasiperiodicity, attractors, repellers, saddles, spiral attractors, or other features of data |
|
The Role of Cause and Effect
Relationships |
Conventional approaches seek to explain cause-and-effect relationships between variables in a controlled setting. The interest in controlling for confounds or covariates demonstrates the assumption that phenomena are multiply determined. |
Dynamic Systems do not rely on establishing cause-and-effect relationships, but rather view variables as not only multiply determined, but mutually interdependent. |
|
What is error in the results? |
ÒNoiseÓ as error |
ÒPerturbations to systemÓ rather than error |
1. Dynamic Systems is not a statistical approach. Dynamic Systems is an approach, a way to re-conceptualize our phenomena.
2. Thus, current statistical programs (e.g. SPSS) which you use and with which you are comfortable are most likely well-equipped for and capable of conducting analyses, creating variables, or making visual representations of data in keeping with Dynamic approaches.
3. You need not re-invent the wheel. Try taking your existing theory or phenomena of interest and imagine how the basic tenets may change or how existing findings may look different when examined dynamically rather than statically.
4. Try applying dynamic systems terminology to your constructs.