The question of whether thought is exclusively verbal or linked to language capacity is not answerable from a phenomenological point of view, since we can think of instances of mental symbol-use that do not rely on language; on the contrary, we know that language “fills in the void”, so to speak: when we acquire language, it fulfills the cognitive demands to express references and relations among them. In this view, thought and thinking precede the linguistic representation of the involved concepts.
If one wants to follow this line of thought, it remains to be shown how the Davidsonian dictum that animals do not have a special form of cognitive ability, namely, propositional attitudes such as beliefs, desires etc., relates to the general argument on higher-cognitive faculties, which do not depend on language possession and which are of the same kind across the animal and human realms. It would thus have to be argued for a language-independent form of propositional attitudes.
Does the inner-speech bias bear not only on thinking at large but also our self-attribution of desires and beliefs? It might. Roskies rightly raises the question, contra Davidson, of whether all our beliefs have definite content (this collection, p. 6). In my view, as soon as we hold a belief qua belief, some kind of cognitive meta-representation must come into play. Such a form of meta-representation strikes me as probably being conveyed by the inner speech mechanism and as thus being subject to the phenomenological inner-speech fallacy.
Roskies nicely disassembles Davidson’s arguments and reconstructs them in a clear and easy-to-follow fashion. She exposes their argumentative weaknesses (such as the issue of interpretation and behavior) and provides ample empirical examples of, and conceptual arguments for, why we should not follow Davidson in his assessment of animals’ cognitive abilities. However, I have tried to show that a further underlying claim can be made, namely that not only is animal cognition a matter of speculation, but that even our own inner workings are less transparent than we commonly like to assume. Davidson’s claim rests, to my mind, on the rashly embraced yet unfounded assumption that language plays a key role in higher cognition in humans (1984, 2001). In my view, contemporary research efforts in the cognitive sciences, but also in philosophy, undermines—or at least calls into question—this assumption. Certainly we are dealing with an important philosophical claim, which could only be properly backed up by extensive empirical evidence pointing to the ubiquitous involvement of language-processing brain areas and mechanisms in higher-level cognitive tasks such as decision-making, action planning, deliberation, etc. Doubtless, human cognition benefits from the linguistic format; abstract thoughts about, e.g., liberty can probably only be executed at a significantly deep level if the relevant concepts have been provided by a linguistic community. But the need to express a certain feeling, like freedom as the opposite of (the feeling of) constraint, for example, certainly originates in a pre-verbal or non-verbal manifestation of this feeling.
Focusing on language, therefore, blocks a fuller examination of what thinking in humans amounts to. We have, I believe, misled ourselves in the face of the phenomenology of inner speech as to what it is like to think, for us as humans. But this gets us only part way towards a full understanding of the underlying mechanisms, structures, and sources of thoughts.