The monkeys in Miyashita's experiments were trained to respond consistently to pairs of visual stimuli generated by a complicated parametric process. Following training, recordings were made from the inferotemporal cortex, while the same stimuli were presented to the animal. As should be expected if the monkey's ability to form associations is anywhere near that of humans, the behavioral response to a stimulus pattern associated with another one during training was that of reliable recognition. Strikingly, the electrophysiological findings paralleled the behavioral results. In addition to neurons selective to either of the paired patterns, neurons were found that commenced firing at the presentation of the first stimulus in a pair and remained active until the second stimulus was shown. This result can be naturally accounted for if one assumes that lateral links are created between neurons that respond selectively to the two members of a pair of stimuli (the adjective ``lateral'' is justified in this case because all the neurons resided in the same area of the cortex, at the same general level in the processing hierarchy).
Up to this point, our discussion of lateral connections has been subject to two qualifications. First, these connections were invoked in a post-hoc fashion, as an explanation of phenomena not directly or necessarily requiring association (such as the formation of RF profiles in the Shaping the RFs section, or mental rotation in the Emergence of ... section). Second, the existence of lateral connections was supported by indirect evidence (such as the anatomical mapping combined with functional imaging in the RF case), or was surmised to facilitate modeling (as in the case of the model of mental rotation based on lateral connections). Miyashita's findings directly corroborate the existence of functional lateral connections, and do so in the context of a task which deals with the formation of abstract (and not merely similarity-based) associations between patterns.