So Many Are “Few,” but so Few Are Also “Few” – Reduced Semantic Flexibility in bvFTD Patients

The processing of quantifier words such as “many” or “few” is a complex operation supported by a plastic fronto-parietal network predominantly in the left hemisphere. The internal reference criterion defining a quantifier (e.g., ≥50% for “many”) can be modified in a learning paradigm. Most interestingly, changing the criterion for one quantifier also leads to a change in the criterion for the untrained quantifier, i.e., a semantic restructuring effect, which is supported by Broca’s region in the left inferior frontal cortex. Here, we applied this paradigm to patients with the behavioral variant of fronto-temporal dementia (bvFTD) because they suffer from loss of cognitive flexibility, reduced ability to process quantities and their values, impaired reinforcement learning, and language comprehension deficits. The question was whether the patients would be able to perform the task, show direct learning of the new quantifier meanings, and exhibit cognitive flexibility in terms of semantic restructuring. Eleven bvFTD patients took part in two behavioral experiments. In Experiment 1, in a first baseline block, each individual’s criterion for “many” and “few” was assessed. In block 2, subjects received feedback about their decisions. Contrary to their initial notion, a proportion of 40% yellow circles was reinforced as “many.” In block 3, the effect of this training on their judgments of “many” and “few” was re-assessed. The group of bvFTD patients showed a learning effect for the new criterion trained for the quantifier “many,” but failed to generalize this criterion shift to the other quantifier “few.” Experiment 2 was similar to Experiment 1, but the patients were trained in Block 2 to judge 60% of circles as “few,” with no training for “many.” Again, there was an average learning effect for the trained quantifier “few” over the entire group, but no generalization to “many.” Since the patients were still able to perform the task and showed learning of “many” to direct feedback, the data suggest that the generalization process, rather than initial learning, is more vulnerable to fronto-temporal degeneration.