Forensic reason

Ordinarily, we aren’t very clear about what we’re talking about when we talk about reason. Hence, rationality in particular gets invoked in largely economic terms (e.g., to behave rationally is to maximize economic self-interest). Hence, reasonability gets used interchangeably with plausibility, albeit with slightly more normative force than the latter term. Hence, reasonableness gets praised as a virtue of the genial. Hence, reason as a process and “reasons for” an action or event get confused. So let’s clarify.

Most generically, reason refers to an interminable, rule-based process of self-revision by means of which possible epistemic updates are proposed, evaluated, and then either discarded or integrated into directed graphs of contingent truths. As indicated by ordinary and specialist languages alike, reason can be formal (inference, logic) or informal (heuristics, instrumental reason), which is not to suggest that formal and informal modes of reasoning are unrelated. Cognate terms like rationality and reasonability refer our attention to the coherence, consistency, or revisability of a given epistemic stance or structure (e.g., of the specific relationship that obtains between a belief and the reasons for a belief).

By “forensic reason,” I mean something specific. The word “forensic” stems from the Latin forēnsis – meaning “of the forum, public” – which itself derives from forum, referring to “an enclosure” (e.g., the Roman forum) in which criminal cases were presented publicly. Hence, the dual connotation of the term. On the one hand, the term “forensics” today (as in “forensic science”) generally connotes the domain of a criminal investigation, especially insofar as such an investigation uses data as evidence in court. On the other hand, older uses of the term refer to “the art or study of argumentation and formal debate.” Each use of the term emphasizes the element of rational construction (e.g., of a well-formed argument) or reconstruction (e.g., of a crime scene) in a publicly accessible fashion. Perhaps an argument and a crime scene are not as dissimilar as they seem.

By forensic reason, I mean the following: Forensic reason consists of both conditionally inferential and contextually inferential processes entailing the construction, elimination, and ranking of hypotheses so as to arrive at an at least heuristically defensible explanation (or array of equally underdetermined explanations). Neither “hypothesis” nor “inference” here are intended in the strict statistical sense; indeed, both terms predate and exceed that sense. Rather, “conditionally inferential and contextually inferential processes” refers to the ad hoc (yet rigorously stated), post-data positing of possible hypotheses, which are calibrated, contrasted, and interrogated ruthlessly. In this context, “hypothesis” can be used elastically, its purview ranging from any part of an explanation to the whole explanation itself.

Consider this toy example. If I want an explanation as to why my refrigerator door is open at midnight, a satisfactory explanation will not relate simply the fact that the door is open because its hinges are angled obtusely. Prima facie, the number of possible hypotheses that might explain why the door is open is indefinitely large: (1) the latch is faulty; (2) my spouse was careless; (3) …; (n) aliens were careless. Without sufficient context (e.g., the presence of evidence indicating aliens), some principle of explanatory parsimony discourages us from adopting (n). Note that the principle of explanatory parsimony has nothing to do with the true explanation (call this תמֶאֱ, or emet, meaning “truth,” composed of the first, middle, and last letters of the Hebrew alphabet). Adopting (n) entails inferential reliance on a number of unwarranted additional hypotheses (e.g., aliens exist, aliens were in my kitchen last night). Those additional hypotheses may or may not be true, but, without sufficient context, they remain extraneous. What qualifies as “sufficient” here varies tremendously. If you live in a universe where aliens are commonplace, then a hypothesis involving aliens already involves fewer additional hypotheses (now you need only to determine whether or not aliens were in your kitchen last night). Barring (n) (and leaving aside (3), i.e., all other possible hypotheses), (1) and (2) remain. They are not mutually exclusive. At this point, further contextual elements (e.g., a loose screw, or the remains of a midnight snack) probably allow me to rank the two hypotheses in some fashion without introducing unwarranted additional hypotheses.

What should stand out about what I am calling forensic reason is that it involves the determinate exercise of both analytic and empirical review (e.g., of evidence, of what hypotheses both imply and presuppose) without either adopting a rigorous experimental framework or reifying any particular methodology. Forensic reason is a process, yes, but it is a heuristic, limited process that intends to formulate and rank a distribution of possible hypotheses about some specific phenomenon. As such, forensic reason is operative only in relation to the specific phenomenon it concerns. This is where forensic reason differs from inference to the best explanation. Rather than trying to describe such inference in general, or even to provide a best explanation, forensic reason is a process seeking to grasp in thought the distribution of more and less likely hypotheses contributing to any eventual explanation, regardless of the degree to which that explanation is best, or the degree to which it approximates the truth. Hence, forensic reason is not only error-tolerant, but integrates error into the distribution of possible hypotheses it seeks to articulate and process. In other words, the point of talking about forensic reason is neither to formulate the best explanation, nor to seek emet, but to understand the branching, ongoing process by means of which possible hypotheses are constructed, eliminated, and ranked.