ISCE  Institute for the Study of Coherence and Emergence

Mechanism versus Category as Explanation:  Spring 2013

A look at how our mode of explanation affects our affordances for action.

Change results from cause, and cause is the subject of explanation. Two types of explanatory models are often evoked as the context underlying change. Models based on labels and categories we shall refer to as “representations.”  More complex models involving stories, multiple algorithms, rules of thumb, questions, ambiguity we shall refer to as “compressions.”  Both compressions and representations are reductions.  But representations are far more reductive than compressions.  Representations can be treated as a set of defined meanings – coherence with regard to a representation is the degree of fidelity between the item in question and the definition of the representation, of the label.  By contrast, compressions contain enough degrees of freedom and ambiguity to allow us to make internal predictions so that we may determine our potential actions in the possibility space. Compressions are explanatory via mechanism.  Representations are explanatory via category.  The danger is when we confuse the evocation of a representation (category inclusion) as the creation of a context of compression (description of mechanism).

"If we wish to understand a real thing, be it natural, social, biosocial, or artificial, we must find out how it works. That is, real things and their changes are explained by unveiling their mechanisms: in this respect, social science does not differ from natural science. Any explanation involving reference to a mechanism may be said to be mechanismic. This qualifier distinguishes explanation proper from mere subsumption of particulars under universals-as in the standard "covering law model" of scientific explanation proposed by the neopositivists. I submit that all concrete systems are endowed with one or more mechanisms that drive or block their transformations. (The rule is one mechanism-one system, not the converse.) Every mechanism is thus a mechanism for either change or control of change. The change may be quantitative, qualitative, or both at once. Mechanisms can be causal, probabilistic, or mixed. Because most mechanisms are hidden, they must be conjectured before they can actually be discovered. The disclosure of a mechanism starts by analyzing the system in question, that is, by showing (or conjecturing) its composition, structure (relations among the parts), and connections with the environment. It proceeds by showing (or hypothesizing) what the system components do (specific function) and how they do it (specific mechanism)." (Mario Bunge)

These ideas were emphasized in a recent work by William Bechtel who notes, “Mechanisms are bounded systems, but ones that are selectively open to their environment and that often interact with and depend upon their environment in giving rise to the phenomenon for which they are responsible. As Robert Cummins (2000) notes, in psychology laws are typically referred to as effects and they typically characterize phenomena in need of explanation but do not themselves explain the phenomenon. Rather, what serves to explain a phenomenon is an account of the mechanism responsible for producing it. A mechanism is a structure performing a function in virtue of its components parts, component operations, and their organization. The orchestrated functioning of the mechanism is responsible for one or more phenomena. (Bechtel & Abrahamsen, 2005; Bechtel, 2006)”

“Mechanisms exist in nature whereas mechanistic explanation involves an investigator presenting an account of the mechanism taken to be responsible for a given phenomenon. Typically, the explanation involves describing or depicting the component parts, operations, and their organization (diagrams are often far more useful than linguistic descriptions for this purpose). The fact that the operations that parts of a mechanism perform are different from the phenomenon exhibited by the whole mechanism and individually do not realize the phenomena makes the working parts of a mechanism different from domain-specific modules. Understanding how the orchestrated operation of the parts produces the phenomenon of interest, investigators must simulate the operation of the mechanism, either mentally or by using model systems or computer simulations.” (Bechtel & Abrahamsen, 2005; Bechtel, 2006)”

Cognitive systems cannot be adequately described by linear models, labels and categories.  This means that the economists’ ceteris paribus is never true.  It means that complex systems cannot be compressed into indexical representations.  The sameness of the indexical is the very absence of context which evokes mechanism.