In Kant's Transcendental Psychology (hereafter, KTP), Patricia Kitcher gives an insightful argument for the inadequacy of the law of association, which she asserts was Hume's primary explanation for how we come to believe in objects (KTP, p. 69):
"The law of association does not explain how we represent objects; the law is itself explained by the rules of synthesis that make reference to objects possible:" (KTP, p. 77)
Her argument is based on the idea that association is dependent on some other source of unity, or on some other determining principle to provide order to "multiple cognitive states from various sensory modalities" (KTP, p. 72). The diversity needs unity, and that diversity can't be the source of that unity. Furthermore, if the data, whose order we are trying to explain, depends on a unifying principle, it is no explanation to appeal to a law which also depends on a unifying principle:
"Without the assumption that ideas have determinate, repeatable contents, the law [of association] makes no sense. In that case, however, the law presupposes and cannot explain our ability to construct representations of objects and properties." (KTP, p. 78)
However, after expressing the argument so well, she said the following:
"However, the fact that this sense-based aspect of mental life is inadequate does not show the impossibility of some other, more subtle pattern of sensory stimulation being the basis of object representation." (KTP, p. 78)
This seems to be inconsistent with her previous argument. As she said herself, "Our senses do not take in whole objects or their properties through some type of migration" (KTP, p. 71). On what basis, then, can she argue the possibility of "subtle patterns" which seemingly could only be understood as properties that somehow migrated? She also rules out the possibility that spatiotemporal contiguity could serve as an explanation, saying, "The stream of cognitive states is too fluctuating and too varied to yield representations by the simple mechanism of spatiotemporal contiguity." (KTP, p. 79) However, in spite of that, she suggests that there might be some other way.
Kitcher goes on to criticize Kant's argument based on the possibility that her hypothetical "subtle patterns" actually exist:
"Because Kant's argument for a priori elements is really an argument against the law of association, it is not totally successful. He does not rule out the possibility just described, so he cannot legitimately claim that our representations of objects do not derive from the senses." (KTP, p. 80)
How can she suggest the possibility that the recognition of these sensory patterns is exempt from the inadequacy of the law of association that she argued so well to establish? According to her argument, this exemption would have to be due to an independence from any organizing principle, as if the patterns could arrive with self-explanatory properties. Is there some loophole to her argument which allows subtle properties to migrate along the neural pathways independent of any need for interpretation?