I continue my series of short comments on Daniel Dennett’s. These comments are mainly written for an online forum, in which undergradute students of Professor Dennett post. Each time I raise a point about one chapter of his recent From Bacteria to Bach and Back. This time I comment on chapter nine.
I took some issue with Professor Dennett calling words minimal agents and comparing them to viruses (p. 189). Allow me to explain what I take to be the decisive difference. Words do not have a structure of functionally differentiated parts. They are not organisms in this sense of the term. They do not have any perceptual organs, nor any parts exhibiting a mechanism analogous to perception. Viruses have at least some functionally differentiated parts. The letter a in the word apple has no corollary differentiated function.
Ultimately, the components of a word lack the relevant causal interrelation. There exists causal interrelation between the parts of an organism, even something as simple as a virus.
Does this rule out that words undergo a process akin to natural selection? The list of features needed for such selection given by Godfrey-Smith (variation, differential reproductive success, and heritability) and endorses by Professor Dennett (p. 138) does not include a structural necessity of functionally differentiated parts. Natural selection might still apply, but the difference makes me wary of drawing too many analogies beyond that. I would resist the temptation to call a word a minimal agent and calling it an agent, even metaphorically speaking, appears misleading to me.
P.S. Historically God has been considered a perfectly undifferentiated agent in theology. I doubt, however, that Professor Dennett wants to put himself in this tradition of thinking about agency.
UPDATE (8. 3. 2017):
I learned from Professor Dennett’s comments that I misunderstood his notion of words. In his terminology, at least as used in From Bacteria to Bach and Back, “words” refers to informational structures instantiated in the brain, rather than to entities composed of letters. (He pointed to Ray Jackendorff’s work.)
I am still not sure, whether we should ascribe the same kind of functionally differentiated parts to informational structures we ascribe to organisms. The issue becomes more difficult, however, than my original comment made it appear.