Some thoughts on a priori, a posteriori and other types of conclusions

Philosophers commonly distinguish between a priori and a posteriori conclusions. The former are conclusions whose truth can be determined solely by virtue of the meanings of the terms that make them up. The common example is the statement all bachelors are unmarried. The philosopher Kant, who articulated this distinction, defined a priori conclusions as those in which the predicate is contained within the subject. By contrast, a posteriori conclusions are those that are true only by virtue of some external facts. For example, the truth of the claim that it will rain in New York on April 4, 2019 can only be ascertained by observing aspects of the world external to that statement.

In the sciences, a related distinction can be seen between theoretical/philosophical work and empirical research. Both lead to conclusions which can be asserted with some certainty (assuming we exclude truly haphazard theorizing). Corresponding to Kant's distinction (above), theoretical/philosophical work involves reaching conclusions that are not immediately dependent on observation, something which is the case for empirical research.

While this distinction has proven to be useful and relevant, I would like to point out how processes of drawing conclusions do not necessarily fall into only one or the other category in a mutually exclusive way--which is not to say that more or less pure cases of drawing conclusions of either type can't be found. 

To illustrate this, I'd like to use a recent project (my dissertation) as an example. My dissertation analyzed the discursive functioning of what I call knowledge claims. Knowledge claims are claims about what people know of the form s/he knows that X, s/he has a concept of Y, or s/he knows how to Z. The goal of my work (which is the subject of a forthcoming article in Theoretical and Philosophical Psychology, co-authored with Shahida Abdulsalam and Eugene Vvedenskiy) was to  understand what it means to know something by analyzing the ways that knowledge claims function in discourse. Specifically, under what circumstances (esp. for what reasons) may they be asserted or contested? Concerning the reasons that are given to support the assertion or contestation of a claim, what is it about these reasons (and not others) that makes them valid?

As the forthcoming article explains, the things that knowledge claims claim that people know (e.g., how to ride a bike, that Albany is the capital of NY, a concept of number) appear to be generalized descriptions of behavior, rather than of any kind of brain or cognitive content. This conclusions was reached through the careful analysis of texts (published research articles on children's understanding of number) in which knowledge claims were extensively used.

Despite the reliance of my analysis on published texts as a type of data source, I would not consider the conclusions to be a posteriori. The reason is that, although the conclusions were informed by the analyzed texts, the specific aspect of the texts that informed the conclusions--and the subject matter of the conclusions themselves--were conventional/normative in relation to both the researchers' and assumed audience's perspective uses of knowledge claims. Therefore, although an external data source was analyzed to inform the ultimate conclusions, the conclusions could have been reached without consulting any external data source. The "researcher" could have just made up the data. In the case of the knowledge claim project, this would involve thinking up various ways that typical interlocutors might assert and justify knowledge claims in conventional ways. The analysis of already-existing texts simply stimulated the analysis towards relevant features of normative discourse that might have otherwise been laborious to think up.

While the preceding argument would seem to suggest that the conclusions about knowledge claims were a priori, this is also problematic. A priori claims are those that are true by virtue of the meanings of the terms themselves--i.e., by virtue of the normative meanings of the words that constitute them. The conclusion that knowledge claims are descriptions of behavior is not self evidently true based on the definition of knowledge claims. (If one objects that knowledge claims are too obscure a category, the same point can be made with a particular example of a knowledge claim: Having a concept of number does not appear, self evidently, to be a description of behavior.)

It might even be argued that it is an a priori truth that a description of what someone knows is not a description of what they can do. Yet, the claim that this is not true is corroborated in a number of different ways, and--at least in principle--this corroboration could have been done exclusively with spontaneously imagined examples of normative conversions involving knowledge claims.

Some musings on the "sameness" of standardized experimental stimuli

The traditional approach to experimental research in psychology involves the creation of experimental settings with standardized stimuli and measures of response. Significant care is taken to eliminate differences in the stimuli presented to participants, except insofar as this constitutes manipulation of the independent variable. For example, a study investigating the effect of the color of light on mood might expose participants to environments that vary in color, but in no other way. It is assumed that participants exposed to the same color environment are being exposed to the same stimulus.

It's not hard to see how this can become problematic. In a psychological context, different qualities (e.g., color) are not readily manipulated independently of other qualities For example, changing a heart from yellow to purple is hardly just a change in color, since the latter color evokes the U.S. military service award the purple heart. Consequently, the variable color is confounded with presence or absence of whatever associations a particular participant may or may not have with the military medal.

There are two issues here. The first is the way that variables that are analytically distinct may be conflated empirically (color with connotations of military service). The second is that the fact that this conflation is not solely a result of the stimulus itself, but also of a particular participant's interpretation of that stimulus. In other words, the "same" stimulus can be interpreted in different ways by different participants (or the same participant on different occasions). In the color research example, this means that some participants exposed to the purple heart would perceive the connotation to military service honor and others would not. In effect, participants ostensibly exposed to the same stimulus have actually been exposed to different percepts.

This second issue--which has been raised by a number of critical commentators--is hardly hidden. It's fairly obvious that different stimuli could have different meanings for different participants. Still, the fact that researchers go on as if stimuli can be treated as "the same" for different participants is hardly just an error.

In acting as if the same stimulus will be "the same" for different participants, researchers are making an assumption that underlies human social interactions in general. The foundation of human cultural life is intersubjectivity, i.e., the assumption of a shared world of objects which exist for me as they do for you. If we were to replace wholesale the assumption of intersubjectivity with the skepticism that things as they appear to me are not necessarily as they appear to you, language would be impossible.

While accepting the basic reasonableness of the critique of standardized stimuli, it's worth considering whether the intersubjective basis of human cultural life entails some important role for standardized stimuli. In other words, what does the centrality of the assumption of a shared environment in human life in general mean for research contexts where this assumption may be made?