Understanding why psychotherapy works in a neuroscientific and mathematical way has been a long-standing interest of mine. Although my idea on this topic is still preliminary and lacks scientific evidence, I’d like to share it here to refine and develop it shortly.

Viewing the brain as an information-processing machine is a well-known approach in cognitive psychology. When considering psychotherapy as an information processing task, from the client’s perspective, it can be seen as having an input, internal processing within the brain, and an output. The brain encodes, stores, and updates information. Additionally, the brain can consciously perceive only a subset of information processes, while the rest remain unconscious. The information, opinions, beliefs, and intentions that enter our consciousness can be seen as a result of competition, each carrying a certain weight that represents its importance in the processing system. Emotions play a role in this weighting process, as information is labeled and the stronger the emotion associated with it, the higher its weight. Moreover, information with a higher weight can have a greater impact on our thinking processes.

In my view, psychotherapy involves identifying high-weight information that unconsciously influences the client in a negative way (similar to bugs) and bringing it to conscious awareness, then modifying, replacing, or eliminating it. It assumes that all information should be conscious to be addressed. In cognitive behavioral therapy (CBT), the process of identifying high-weight information can be the conscious recognition of hot cognitions or irrational beliefs. High-weight information can be brought to conscious awareness through recall, enabling modification, and reducing weight. For example, it entails the reconsolidation of traumatic memories to alleviate the intense fear attached to them.

Considering psychotherapy as a work of two information processing machines (therapist and client), how can we understand the important warm elements in psychotherapy such as rapport? In my view, the rapport between the therapist and the client can be seen as a lubricant that enables the therapist to access and organize information from the client’s inner world. In other words, rapport might be a way to make the client more receptive to the therapist’s intervention.

Then, if the client is receptive to any intervention with or without lubricants like rapport, could AI psychotherapy also be effective? It seems plausible. However, it may not be easy to reach a level of human therapists. They capture the moment-to-moment responses of clients while taking into account the psychological state of each individual. Simply confronting the psychological issue may not be enough to make the client voluntarily recognize and modify their unconscious inadaptive thoughts (information). The therapist needs to react promptly and appropriately to each moment, helping the client organize and reflect on the information arising from within.

So how can we scientifically validate this perspective? In my field of study—computational modeling of aberrant decision-making patterns in people with psychopathology—there are recent attempts to capture changes in decision-making patterns before and after or across psychotherapy sessions. However, as my idea here focused on identifying and modifying implicitly held high-weight information, decision-making patterns (e.g., risk-taking, learning rate) are less relevant. Rather, the patterns seem to be related to individual differences in the reception of information or the speed of updating.

Another approach could involve examining the activation of brain regions during the psychotherapy session, those related to turning implicit information into explicit information. However, to do so, we would need to understand the neuroscientific mechanisms underlying the transformation of implicit information into explicit information, as well as non-invasive tools for observing this process in humans.

The idea presented here is very preliminary and also does not encompass various concepts used in psychotherapy. For instance, how could ‘emotional awareness’ be understood? Nonetheless, I hope that it might be a baby step toward establishing a framework for scientifically validating the mechanism of effective psychotherapy.