We were having our daily coffee, and my friend did it again: She asked the waitress if her coffee could be in a glass mug or cup; anything made of glass, but never a regular mug. After observing this behavior day after day, I finally confronted her: “Does it make that much of a difference?” She looked at me, puzzled: “What? The glass? Absolutely! I can’t drink coffee in a regular mug.” I was baffled. To me, coffee’s tastiness is vessel-independent. For her, mug and glass mug are two absolutely different things, almost incommensurable. How is that possible?

Our survival is possible because every day we make use of previously acquired categories to navigate the world. Every single mug we encounter is distinct, but fundamentally the same. Thanks to this powerful capacity to classify distinct stimuli under the same category, we can generalize our knowledge from the previously encountered subset of mugs to a future subset of mugs. However, this also posits a dilemma: Is a glass mug still a mug? That is, what are the defining principles that make something a “mug”? Establishing this is fundamental since it also affects its relationship with its close-neighbors. Conceptualizing a mug as very different from a glass creates a more clear-cut mapping between the input—that is, the stimulus perceived—and the output that a person needs to generate—that is, the response, such as drinking coffee. Classical work in cognitive science demonstrates that the more similar two stimuli are, the harder it is to discriminate them and respond with different behavior. Building on work in individual differences, we hypothesized that people averse to uncertainty will be especially drawn to conceptualize stimuli more dissimilarly. By expanding their semantic representations, they can reduce uncertainty originated from having too many concepts activated at a given time: If one can only drink coffee in a glass mug, that is the only concept activated in that context, and no other alternatives are considered.  

To test this semantic-expansion hypothesis, we first ran an experiment in which we measured how people conceptualized 16 words, such as freedom, hospital, children, or family. The prediction was that those averse to uncertainty will tend to, overall, represent these concepts as more distinct than people tolerant of uncertainty. Our prediction was confirmed: The average psychological distance between concepts increased with aversion to uncertainty. We then moved to test the same hypothesis at the neural level: Could it be that the neural activity patterns associated with words are, on average, more distinct for people averse to uncertainty? That is, the neural activity pattern of thinking “mug” is more dissimilar than the neural activity patterns associated of thinking “glass”. We obtained evidence for our prediction in the left inferior frontal gyrus, a region classically related with the resolution of semantic competition (see figure below). Across two experiments using distinct methods, we found evidence that people averse to uncertainty have a more expanded semantic representation at the psychological and neural level.

But what about the behavioral consequences of this semantic expansion? Is it true that it prevents confusion? We ran a new experiment to answer this question. In this case, participants were given a surprise recognition test in which they had to classify a set of words as “old” or “new” depending on whether they had seen them before during an unrelated task. The probability of recognizing a “new” word as “old” increased with the semantic similarity between the words. That is, if an old word is “white” and a new word is “snow”, it is more likely to falsely recognize “snow” as old than if the new word was “carpet”. This effect occurs because when thinking about white, there is a co-activation of related concepts like snow. However, if you are uncertainty averse and have a more expanded semantic representation, then that effect should be alleviated, because each concept is more separated from the other, which should decrease the amount of co-activation. Therefore, as a consequence, “snow” should be more protected from confusion with the previous presence of “white”. And that is exactly what we found: People averse to uncertainty were less affected by the semantic similarity between the old and the new words. After all, the semantic expansion might protect them against confusion and aid discrimination.

Different personality needs cause a distinct organization of people’s semantic representations, which facilitates discrimination. But what about the downsides? It seems that people averse to uncertainty possess a sort of superpower, which allows them to discriminate better, with no costs. Alas, as Borges taught us: “To think is to generalize, abstract, to forget about differences”. That is, a more expanded semantic representation predicts, at the same time, a decrease in generalization. And that is what we went on to test in the last experiment. In this case, we used pictures representing concrete concepts (e.g., “chair”) and participants learned the relationship between a set of pictures and an action (pressing a key). After they learned these concept-action associations very well, they were presented with new images of different concepts, and they had to decide whether they wanted to press the same keys as before or not. Pressing the same key would be evidence for generalization. And here, people averse to uncertainty struggle. They generalized less. Importantly, they generalized less especially for very similar concepts, like “chair” and “stool”. For people averse to uncertainty, what is learned about a chair does not apply as much to a stool.

The study of how the mind works has historically disregarded individual differences. After all, what makes us human, unique, and similar is more important than what tease us apart. Or so the story goes. But, by looking at the powerful effects of personality needs, we can also uncover that even the most fundamental aspects of our cognition, like the structure of our semantic representations, is affected by who we are and what we need, such as reducing uncertainty or choosing a simple coffee in a glass.