We Met Yesterday, Will I Like You in Three Months?

Neuroscience research on how brain activity predicts future mutual affection.

Posted Apr 16, 2018

You can't find true affection in Hollywood because everyone does the fake affection so well.

—Carrie Fisher

The heart hath reasons

Whenever we meet someone, we have some feelings about whether we like the other person, and if so, how much. Sometimes it’s a dud, no chemistry. Sometimes it’s nice, you like them but you aren’t bowled over. Sometimes, it’s like a jolt, and you immediately are very drawn to the other person. This immediate attraction doesn’t have to be romantic. In fact, it is quite common in many relationships—meeting a potentially great future friend, being introduced to a mentor figure, meeting your child’s new teacher, and so on. As long as there is a reasonable chance of a significant relationship in the future, we will notice the feeling of affection.

Liking and being liked usually feels really good

What’s going on here? As Columbia University researchers Zerubavel, Hoffman, Reich, Ochsner, and Bearman (2018) discuss in the recent research described below, as with other of life’s enjoyable experiences, when we meet someone we like, reward networks in the brain become more active. Specifically, they report that ample research has demonstrated that both the ventromedial prefrontal cortex (vmPFC) and the ventral striatum (VS) become more active when social rewards are involved.  

The vmPFC, evolutionarily an area of higher brain development, has many functions. Among others, the vmPFC is involved in social decision-making, weighing alternatives to contribute to good choices. People with damage to this area may have difficulty detecting irony and sarcasm, for example, and may be more easily swayed by misleading advertising. The VS, a deeper, older part of the brain, is involved with risk and reward, learned behavior and habits via reinforcement by reward, and in extreme cases obsessions and compulsions, and addictive behaviors. The VS is connected with basic operational areas in the forebrain, underneath the outer cortex where the vmPFC and other areas are.

It makes sense that the vmPFC and VS are active in tandem when we are with someone we like, because the VS is active whenever anything is rewarding, and the vmPFC is, in a sense, calling the shots on who is determined to be enjoyable to be with, and who is not.

Of course, the other person is doing the same thing you are doing when you meet, sizing you up in many ways, including seeing whether they like you, how much, in what ways, and whether you like them back. We pay attention to how much we think the other person likes us and that is part of how much we like them. When we know that the affection is mutual, it adds to the intensity of the experience, a feedback loop called “affective” (emotional) or “dyadic” reciprocity. The more we like someone, the more they like us, and the more we like them, and the more we like each other—though there are notable exceptions, such as when liking or being liked is anxiety-provoking and leads to withdrawal from relationships.

The brain activity of shared affection

Zerobavel et al., 2018
Source: Zerobavel et al., 2018

Based on these observations, Zerubavel and colleagues designed a study to see whether they could predict, based on brain activity, whether two people meeting for the first time will end up liking one another several months later, and whether predictive factors are inherently interpersonal, or solely a function of individual (intrapersonal) factors. Researchers were specifically measuring "liking", as contrasted with distinct but important constructs such as trust, influence, and other factors relevant for affiliation [N.B. thanks to Peter Bearman for clarifying this point].

Focusing the vmPFC and VS, in addition to looking at basic brain activity in this social-reward-and-decision-making loop as a predictor of liking, they also analyzed neuroimaging results to see how relationship-level factors influenced mutual affection over time. For measuring mutual affection, there should be relationship-level factors which are independent of individual brain activity.

Studying a group of 16 college students working intensively together on a 9-week summer project, researchers measured their brain activity during a simulated social encounter at the beginning of (T1) and after months of actually getting to know one another (T2). While looking at photos of one another’s faces, fMRI was used to measure activity in the vmPFC and VS. Researchers used structural equation modeling to apply the “actor-partner interdependence model” to fMRI data, allowing researchers to check not only what individual factors predicted mutual affection at the end of the summer, but also whether aspects of shared brain activity were uniquely predictive of future affection. The nature of this affection, romantic, platonic, or what have you, was not specified.

There were several relevant findings. First, they learned that individual brain activity in the areas of interest (“neural valuation”) at T1 predicted feelings of liking the other person at T2, even after factoring out how much they said they liked the person at T1. This suggests a specific neural predictive element beyond how much one person subjectively was aware of how much they liked the other person. They also found that how much one person ended up liking the other person was also correlated with how much the other person liked them at the beginning.

Persuasively, they showed that social factors, like personality and demographic factors (e.g. similarity of background), did not significantly contribute to predicting future liking, underlining the predictive value of neural activity. Researchers were also able to show that both individual factors (“actor effects”) and relationship factors (“partner effects”) were significant in understanding and predicting future mutual affection. They took pains to make sure they were showing a real partner effect, for example testing to make sure that one person’s liking of the other grew over time, above and beyond their own initial feelings of affection and in interaction with the other person’s liking of them. Finally, the importance of the partner effects doubled over the course of the summer, surpassing the early individual markers of affection.

Two hearts beat as one

The study authors elegantly demonstrate that mutual affection is a unique property of the relationship, based on shared experiences of mutual affection which grow and deepen over time as a result of unique interactional factors. In a sense, two minds come to think as one, and moreover, the relationship takes on a life of its own as the whole becomes greater than the sum of its parts. Mutual likeability is of such great importance in any well-functioning relationship, whether it is friendship, professional, romantic, or family, and the research reported here advances our understanding by identifying specific dyadic factors.

Future research could look at a broader range of relationship types and duration, more detailed data sets, a greater number of time points to understand how mutual affection evolves over time under different circumstances—and what happens when we meet someone we markedly do not like, or when we shy away from someone we do like, and who likes us. The general approach to studying affiliation used in this study has great potential for shedding light on many aspects of how relationships succeed and fail. The study of “neuro relationship” holds promise for enriching understanding of our inherently relational human experience.

What additional subjective factors would correlate with predictive neural valuations corresponding to brain activity in vmPFC and VS? Future research could also look at narrative descriptions and an expanded set of measures to derive a more sophisticated predictive model, perhaps looking at activity across the whole brain using machine-learning for a much larger dataset. It may be a while before this technology makes it to dating services and HR departments, but it presents intriguing possibilities. Perhaps even telling us whether affection is genuine...

In the meantime, next time you meet someone new, remember that your vmPFC and VS are busily humming along in the background, not only for you—not only for the person you are meeting—but also in ways unique to the two of you together, transcending either of you as individuals.


Zerubavel N, Hoffman MA, Reich A, Ochsner KN & Bearman P. (2018). Neural precursors of future liking and affective reciprocity. PNAS April 9, 2018. 201802176; published ahead of print April 9, 2018. https://doi.org/10.1073/pnas.1802176115