Can we learn from our mistakes?
Have you ever made a decision with the best intentions, only to end up with an outcome you later regretted?
The wrong financial investment, the wrong partner, the wrong timing, or the wrong career choice. We often say that people should learn from their mistakes. Usually, this means understanding what went wrong so that next time, you will make a different decision.
But what if you did learn from the experience and still ended up with a similar outcome? For example, you might choose a very different partner and genuinely believe that you have learned from your previous relationship but fear of rejection may still influence what you tolerate, what you avoid discussing, and how long you stay when problems appear. The person may be different and your choices may look different, but the same underlying response can continue to shape the outcome. This is why understanding what happened may not always be enough. We may also need to understand the influences behind our decisions and behaviour before the outcome.
Understanding these underlying influences has been one of psychology’s longstanding challenges.
For most of human history, people relied largely on reflection and observation to explain human behaviour. During the 19th century, psychology began developing scientific methods that allowed researchers to test explanations against evidence and identify patterns across people. Over time, this created an increasingly strong scientific understanding of human behaviour.
As psychological knowledge expanded, researchers could identify common tendencies across populations, while practitioners needed to determine how those findings applied to a particular person and their circumstances. By the 20th century, psychological assessment, structured observation, and evidence based practice provided increasingly systematic ways to understand an individual using knowledge developed through broader research.
Today, researchers continue developing increasingly sophisticated ways to understand how psychological processes develop within an individual across different situations and over time. However, collecting and interpreting enough information from everyday life to build a detailed and reliable picture of one person remains difficult. A psychologist may understand that fear of rejection can influence behaviour, for example, but identifying exactly when that response appears in one particular person, when it does not, what activates it, and what follows requires information across many situations and over time.
Psychology is already trying to address this limitation by collecting information from individuals more systematically and repeatedly. In 2025, the American Psychological Association approved professional practice guidelines for measurement based care, encouraging psychologists to collect information at regular points throughout care so that changes can be monitored more systematically and decisions about support can reflect current evidence.
This represents an important move toward a more detailed understanding of the individual, although repeated measurement still does not always explain the process behind a change. An anxiety score may show that anxiety has increased without revealing what happened beforehand, what the situation meant to the person, how that meaning influenced the response, or what happened afterward.
Artificial intelligence may extend this work by making it possible to compare many more experiences from the same person and examine whether the same explanation continues to fit across different situations. Psychologists already identify behavioural patterns, but the amount of everyday information that can realistically be compared has traditionally been limited, while AI may be able to examine a much larger behavioural history that includes both the situations in which an expected response appears and those in which it does not.
Consider the earlier example of fear of rejection. If that appears to explain someone’s difficulties in relationships, AI could compare many situations in which rejection was possible and determine whether the same response emerged under all of those conditions. It may become clear that the response appears mainly when the person places particular importance on the relationship or believes that losing it would affect their sense of acceptance.
That additional information can make the explanation more precise because the problem may not involve fear of rejection in general, but a more specific response that becomes active under particular personal conditions. This also helps explain why learning from one mistake may not prevent a similar outcome when the deeper influence behind the decision remains unclear.
AI could potentially help identify that process by comparing experiences over time and testing whether an explanation remains consistent with new evidence. This would not replace psychological knowledge or professional judgement, but it could provide a much richer source of individual information for both.
If this approach proves reliable, the practical benefit could be much more direct. Before entering another relationship, accepting another job, or making another important decision, a person could have a clearer understanding of the situations that tend to influence their judgement, the responses that repeatedly create problems, and the conditions under which they make better choices. Learning from experience would then involve more than remembering the last mistake, because the person could recognize what is happening earlier and make the next decision with a better understanding of themselves.
#FELIXA #ArtificialIntelligence #Psychology #BehaviouralScience #AI #HumanBehaviour #DecisionMaking #SelfAwareness #BehaviouralIntelligence