The associative process by which our thoughts follow one another, has intrigued scholars and psychotherapists for decades. Yet, the question of how we can, relatively quickly, choose one association among countless possibilities remains poorly understood.

In a new study, Dr. Isaac Fradkin and Dr. Eran Eldar investigated this question by conceptualizing free association as the accumulation of internal evidence. One mechanism, implemented in neural network models, involves rich-get-richer dynamics – every time an association receives some evidence, the probability that it will continue receiving evidence increases. Thus, the number of associations that remain `in the race` decreases over time. However, rich-get-richer dynamics also mean that weak associations (e.g., `Table` -> `Sky`) should be reported as quickly as strong associations (e.g., `Table` -> `Chair`). Thus, an additional mechanism is required to explain why strong associations actually tend to be faster.

The study has shown stronger associations are reported faster, even under rich-get-richer dynamics, when assuming that the brain represents stronger associations (`Chair`) more similarly to the cue (`Table`). Importantly, these two assumptions are mutually co-dependent: ‘overlapping representation’ without rich-get-richer dynamics leads to overly unimaginative and not creative associations.

The study also investigated the benefits and limitations of alternative mechanisms explaining free association – for example, the idea that activation is distributed simultaneously from the cue to all associations.

The findings and methods developed in the paper can help understand individual differences in free association and thought dynamics, for example, in creativity or some psychiatric dimensions.

See full artice here