The neural segregation of declarative and procedural memory is based on an outdated idea that the mind is separate from the body [but for a recent example see Fig. 3 of Sendhilnathan, Goldberg et al. 2020a]. Both the neocortex and cerebellum participate in the creation of declarative and procedural memories during learning (Tehovnik, Hasanbegović, Chen 2024): the neocortex via the hippocampus stores the declarative/procedural memories (as conscious memories) in the association areas (e.g., the retrosplenial, parietal, infratemporal, and orbito-frontal cortices) and the association areas transfer the declarative/procedural information to the cerebellum for storage in the form of executable code (as unconscious memories) for the rapid execution of a learned behavior using a rate code, which is the language of the muscles (Tehovnik, Patel, Tolias et al. 2021). The neocortex and the cerebellum have an extraordinary capacity for the storage of information (either declaratively or procedurally) estimated to be 1.6 x 10^14 bits (2^1.6 x 10^14 possibilities) and 2.8 x 10^14 bits (2^2 x 10^14 possibilities), respectively, and the individual elements of a memory are concatenated (both in the neocortex and cerebellum) by linking the elements to generate a sequence of movements or words or a string of notes (Tehovnik, Hasanbegović, Chen 2024).
The evidence that declarative and procedural memories are processed together comes from the following experiments. As early as 1927, Pavlov showed in dogs that if he ablated the neocortex that the animals failed to learn new associations using the classical conditioning paradigm, which has been associated with procedural learning. This at first might appear to contradict the findings of Mauk and Thompson (1987) but in their conditioning experiments (on rabbits) transection of the cerebellum from the neocortex abolished previously learned associations only. They never tested the acquisition of new associations.
Indeed, the conditioning studies of Takahara et al. (2003) are instructive in this regard. They tested rats on a conditioning task in which ablations were done early or late in training (i.e., on day one after conditioning or on day 30 after conditioning). It was found that lesions of the cerebellum that included the cerebellar nuclei abolished an animal’s ability to learn the task early as well as late in training. When the same test was done with hippocampal or neocortical ablations it was found that hippocampal ablations abolished early learning but not late learning, and that neocortical ablations spared early learning but abolished late learning. The hippocampal/neocortical findings concur with the accepted view today.
Some might argue that classical conditioning is a highly non-cognitive task and therefore cannot be generalized to cognitive and other task types. It has been found that an object association task (by monkeys) was affected early in learning but not late in learning if the cerebellar cortex was lesioned while sparing the cerebellar nuclei (Sendhilnathan and Goldberg 2020b; also see Ignashchenkova, Thier et al. 2009). A similar result was found for adapting VOR (the vestibulo-ocular reflex) to a prism (by cats, Kassardjian et al. 2005): early learning but not late learning was affected by a cerebellar cortex lesion while sparing the cerebellar nuclei. It is speculated that late learning is mediated by the cerebellar nuclei (Kassardjian et al. 2005), which is supported by the findings of Takahara et al. (2003).
Finally, there is overwhelming evidence that declarative learning, as well as procedural learning (as just mentioned: Takahara et al. 2003), is abolished in the short-term once the hippocampus is lesioned, which prevents information from being consolidated at the level of the neocortex for long-term storage (Corkin 2002; Knecht 2004; Morrison and Hof 1997; Munoz-Lopez et al. 2010; Roux et al. 2021; Scoville and Milner 1957; Squire et al. 2001; Squire and Zola-Morgan 1991; Xu et al. 2016).
Just how far the foregoing scheme can be generalized across different vertebrates, all of which contain a neocortex (or its homologue) and a cerebellum is not known. Nevertheless, we now expect that these structures subserve both declarative and procedural learning (with no mind-body duality); the neocortex manages the storage of explicit information (i.e., of objects, words, odors, tastes, and so on) and the cerebellum manages the storage of implicit information for the grasping of objects, pronunciation of words, and for responding to odors and tastes (i.e., during eating, drinking, fleeing, fighting, or mating).