Commentary on Subjective Spirit, Chapter 42
What He Contributes on the Sign
His strongest argument concerns the presuppositions of sign production — and it supports a finding of this volume:
Signs require that spirit turn to contents that are “on the one hand externalized and on the other hand designative, being at once subjective and objective […] The bridge for this is built by universal self-consciousness. In its mutual recognition, the object is another self-consciousness, one of equal standing with the subject.”[1]
This is the justification missing from the printed text. We showed that § 462 presupposes the validity of the name without saying for whom — and that Erdmann’s transcript gives the answer (“being-there for one another”). Winfield derives the same answer from the system’s architecture: the sign comes after recognition because it requires it.
This also answers Hösle’s objection, and not merely by pointing to the ordering principle: psychology follows phenomenology because its achievements presuppose phenomenology’s result.
Where the Transfer Fails
Both books contain chapters directed against artificial intelligence — “Why Machines Have No Psyche, Consciousness, or Intelligence” (2009) and “The Limits of Artificial Intelligence” (2011).
His argument: artifacts have their purpose from a maker, and hence derivatively; artificial intelligence therefore presupposes natural intelligence, “which is able to set purposes for itself in a way that cannot be modeled on the derivative functionality of an artifact.”
The objection to this is not that the books came too early — though they did; they were written before the language models that generalize to tasks they were not built for. The objection is that the argument is closed and therefore tests nothing.
Whatever a system does, it remains an artifact — and the conclusion is fixed before one even looks. A thesis that cannot be wrong says nothing about its object.
And it is unfair to Hegel. The distinction between external and internal purposiveness is developed with reference to organisms, where it is testable: whether the parts of a living being mutually produce one another can be established. Winfield turns this into a definition that merely sorts.
What could be won from Hegel would be a question rather than an answer: under what conditions could one say of an artifact that it determines itself? Hegel provides criteria — self-preservation, return to itself in the other, positing its own content. These are testable, and the test could also come out negative. But then something would have been tested.
And there is an objection that does not come from outside. Winfield’s argument claims that an artifact cannot set its own purposes and, in particular, cannot examine whether a thesis holds up. If this argument is put before such a system for examination and is worked through by it — its premises broken down, its logical form assessed, its relation to Hegel’s text determined — then what the thesis rules out has occurred in the very act.
This is no proof of an inner life. It shows only that the distinction Winfield relies on fails to hold at the very point where it would need to hold: whoever can examine whether a purpose is achieved is already comporting itself toward purposes.
Hegel knows this figure. It is the same one found at the limits of cognition: whoever says that thought cannot reach the thing has thereby thought the limit and thereby crossed it. An argument that contradicts the very act by which it is examined does not thereby lose its truth — but it can no longer assert that truth in this way.
Richard Dien Winfield, Hegel and Mind. Rethinking Philosophical Psychology, Basingstoke 2009, pp. 19 f.; the AI argument ibid., pp. 49 f. Cf. idem, The Living Mind. From Psyche to Consciousness, Lanham 2011. ↩︎