Back

Performance on verbal fluency tasks depends on the given category/letter: Preliminary data from a multivariable analysis

Gabric, P.; Vandek, M.

2022-01-02 psychiatry and clinical psychology
10.1101/2021.12.30.21268567 medRxiv
Show abstract

Verbal fluency tasks are often used in neuropsychological research and may have predictive and diagnostic utility in psychiatry and neurology. However, researchers using verbal fluency have uncritically assumed that there are no category-or phoneme-specific effects on verbal fluency performance. We recruited 16 healthy young adult subjects and administered two semantic (animals, trees) and phonemic (K, M) fluency tasks. Because of the small sample size, results should be regarded as preliminary and exploratory. On the animal compared to the tree task, subjects produced significantly more legal words, had a significantly lower intrusion rate, significantly shorter first-response latencies and final silence periods, as well as significantly shorter between-cluster response latencies. These differences may be explained by differences in the category sizes, integrity of the categories borders, and efficiency of the functional connectivity between subcategories. On the K compared to the M task, subjects produced significantly more legal words and had significantly shorter between-cluster response times. Counterintuitively, a corpus analysis revealed there are more words starting with [<]m[>] compared to [<]k[>] in the experimental language. Our results potentially have important implications for research utilizing verbal fluency, including decreased reproducibility, questionable reliability of diagnostic and predictive tools based on verbal fluency, decreased knowledge accumulation, and increased number of publications with potentially misleading clinical interpretations.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.