Addiction May Be as Simple as ‘One-Shot Learning’ Gone Terribly Awry

Groundbreaking research by California Institute of Technology (Caltech) scientists regarding how the brain learns quickly—a process called “one-shot learning”—has turned up insights that may help further explain addiction and dual diagnoses and how treatment can help. “One-shot learning,” as opposed to more gradual, incremental learning over time, happens when we immediately associate a specific stimulus with a particular outcome. For a long time science has not been able to explain how one-shot learning occurs or where it happens in the brain. Recently though, a team of Caltech researchers discovered both the “where” and the “how” of one-shot learning, and their findings, published in the April 28 issue of the journal PLOS Biology, contain potential implications for recovery from addiction and other mental disorders.

How and Where Does One-Shot Learning Occur in the Brain?

First, the “where.” It turns out that immediate associations between a stimulus and outcome (a.k.a. one-shot learning) take place in a part of the prefrontal cortex known as the ventrolateral prefrontal cortex (VLPFC). (The prefrontal cortex itself is the large area of the brain just under the forehead that houses both complex cognitive processes, such as learning, as well as many of the neurological mysteries behind a variety of psychiatric disorders, from schizophrenia, attention-deficit hyperactivity disorder, autism and depression to drug and alcohol addiction.) Using brain imaging while asking test subjects to undertake simple tasks, researchers were able to determine that causal processing happens when the VLPFC couples with the hippocampus (a part of the brain involved in the formation of memories) to turn on a neurological switch for one-shot learning. Next, the “how.” The Caltech team also discovered that the greater the uncertainty about whether a specific stimulus does in fact cause a particular outcome, the quicker the learning. In other words, the lead author of the Caltech study, Dr. Sang Wan Lee, PhD, a postdoctoral scholar in neuroscience, explained, “If you are really uncertain about whether a particular event (outcome) is caused by a preceding event (cue), then you are more likely to associate them together,” in contrast to a “situation where you are more certain about what the relationship (or absence of relationship) is between the two events.” But, wait: Doesn’t it seem logical to assume that learning would happen more quickly when there’s greater certainty of causation between a specific stimulus and a particular outcome? Caltech professor of experimental psychology Shinsuke Shimojo, PhD, a co-author of the study, and Dr. Lee answered that question using the following example:

Imagine you go out to a restaurant and you have a plate of raw seafood (which you have never eaten before), and you also have a plate of chicken, which you have eaten many times before in the past. Later that evening your stomach becomes upset. You then have to work out which of those foods caused you to feel unwell. Because you have eaten chicken a lot before and not gotten ill, you are much more certain about the relationship between eating chicken and illness than the relationship between eating seafood and illness. Thus, you are more likely to infer a causal relationship between eating the seafood and getting ill.

One-shot learning is thus especially useful for survival. Consider the animal whose survival depends on avoiding poisonous plants and berries, Dr. Shimojo said. “The animal sometimes correctly attributes the sickness to the poison they had 24 hours ago (meanwhile they eat and drink other things a lot). So how could they do this? The key is again causal uncertainty. For their ordinary drink and food, they know they are pretty safe (causal uncertainty is very low), except for the rare food with poison.”

How One-Shot Learning Insights May Further Explain Addiction

If such fast-track learning is critical to survival, it can also go terribly awry—which may further explain certain facets of addiction and dual diagnoses and why they are so hard to recover from. One of these is causal inference between a particular cue (like, for instance, an especially stressful event) and its outcome (the drinking or drug use itself). The more novel or unique the cue, and the more novel or unique the outcome, the greater the chance will be that you link them together. And the more likely your future behavior will register and encode that lesson. While the Caltech study did not directly address those brain regions specifically associated with addiction, Dr. Lee was able to offer the following contextualization, with the caveat that his reasoning is currently speculative in the absence of further research: “One situation we can imagine in the context of one-shot learning is that you had a drink/drug for the first time (a novel outcome) after having a very bad day (an unusual event equals a novel cue) and it felt really good; then perhaps you suddenly have an obsessive thought that it’s the drink or drug that makes you get over a bad day.” In this way, the brain can quickly “learn” obsessive, compulsive behavior in relation to a particular substance, so “unlearning” such lessons — which can become wrapped up in a person’s very sense of survival — is how addiction treatment can help, and why treatment is so critical to recovery. Although these latest findings about the where and how of one-shot learning have yet to be applied to addiction treatment, Dr. Lee has some ideas about how they could be. “If the addiction is attributable to an obsessive thought that associates drug/drink (outcome) with a particular cue,” Dr. Lee said, “one might want to try something like cognitive behavioral therapy (CBT), in which patients are repeatedly exposed to another type of outcome in order to quickly weaken the link between the cue and the outcome.” In other words, CBT and variations of CBT such as dialectical behavioral therapy (DBT) may help to deprogram one-shot learning that has gone wrong. Another way treatment can help is through noninvasive brain stimulation techniques such as transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (tDCS), in which magnetic fields or electrical currents restore “neural stability” (Dr. Lee’s expression) to the VLPFC (ventrolateral prefrontal cortex). Such methods may help facilitate new and more effective learning, which aids the process of recovering from an addiction, especially when that addiction accompanies a dual diagnosis that involves delusional thinking. Certain psychiatric disorders that involve delusional thinking, for instance, “might involve a malfunction of our ability to attribute cause and effect,” Dr. Lee said. Treatments that stimulate the VLPFC’s capacity to make more reliable cause-and-effect connections thus can help treat at least some of the dual diagnoses that often accompany addiction. Sources:

  • “Switching on one-shot learning in the brain,” Science Daily
  • “Dysfunction of the prefrontal cortex in addiction: neuroimaging findings and clinical implications,” Nature
  • “Imaging the Addicted Human Brain,” Science & Practice Perspectives
  • “Pinpointing the Brain’s Arbitrator,” Caltech

By Kristina Robb-Dover, M.Div. Follow Kristina on Twitter @saintplussinner

Scroll to Top