Brain Scans Outperform Behavior in Predicting What Markets Really Want

New research suggests that our brains may reveal more about collective market behavior than our individual choices do—even with surprisingly small sample sizes.

Scientists have discovered that neural activity from just a couple dozen people can forecast the preferences of thousands more effectively than watching what those same people actually choose, according to a study published February 25 in PNAS Nexus.

Researchers led by Alexander Genevsky, Lester C. Tong, and Brian Knutson found that brain scans measuring early emotional responses generalize across diverse populations, while behavioral choices proved less reliable when forecasting groups with different demographics.

The field of “neuroforecasting”—using brain activity to predict market outcomes—has been growing over the past decade, but until now, researchers weren’t sure exactly how or why it works.

“One mechanism underlying successful neuroforecasting involves generalizable components of choice revealed by neural measures,” the researchers write. “In contrast, self-reported responses and observed behavioral choices incorporate idiosyncratic preferences which can contribute to accurate predictions of individual behavior but diminish the accuracy of aggregate forecasts.”

Across two separate experiments involving crowdfunding campaigns and online videos, the team compared the accuracy of forecasts based on traditional behavioral measures versus those based on brain activity.

In the first experiment, 37 participants viewed 36 documentary film projects from Kickstarter while being scanned with functional magnetic resonance imaging (fMRI). They made decisions about whether to fund each project, with one random choice becoming a real monetary contribution. These same projects were then shown to nearly 3,000 people online who indicated their preferences.

A similar setup was used for the second experiment, where 40 laboratory participants watched 32 short videos from YouTube’s Discovery and Animal Planet channels. The subjects could choose to skip videos after watching for a minimum time. Almost 1,000 online participants then viewed the same videos and decided whether to continue watching after the first 15 seconds.

The researchers were particularly interested in two brain regions: the Nucleus Accumbens (NAcc), associated with positive emotional responses, and the Medial Prefrontal Cortex (MPFC), involved in integrating various considerations into decision-making.

Surprisingly, only activity in the NAcc significantly predicted aggregate choices across both internet samples. Even more remarkably, when the researchers divided the internet samples into quartiles based on demographic similarity to the laboratory group, behavioral choices lost their predictive power for less demographically similar groups—but NAcc activity remained a reliable predictor regardless of demographic match.

“While counterintuitive, the superior generalizability of brain activity vs. behavioral choice suggests that early anticipatory affective responses might generalize more broadly than subsequent integrative or behavioral responses,” the researchers note.

This finding challenges conventional wisdom that representative samples are always necessary for accurate market forecasts. It suggests that certain emotional components involved in decision-making are widely shared across diverse populations, while the final choices people make are more influenced by individual factors.

The study’s implications extend beyond academic interest. Companies spend millions on market research and consumer testing, often striving for representative samples that mirror their target markets. This research suggests that a small group of brain-scanned individuals might provide equally valuable insights, especially for products and experiences that evoke emotional responses.

“These findings illuminate how brain measures may reveal seeds of choice that can generalize to forecast market behavior—even when behavioral measures cannot,” the authors conclude.

The research also addresses a practical concern: the cost of collecting brain data. Through statistical analyses, the team found that reliable neural forecasts could be derived from relatively small groups—as few as 14-25 participants. While still requiring specialized equipment, this finding suggests neuroforecasting might be more accessible than previously thought.

According to the researchers, “When a representative sample is available, the addition of neural data might still account for added variance in forecasts of aggregate choice.” Even more valuable, “when a representative sample is not available (e.g., due to lack of accessibility, unwillingness to participate, or missing information about the target audience), brain measures might still effectively forecast aggregate choice.”

The study builds on a growing body of research showing that neural data can improve forecasts across diverse markets, including music sales, crowdfunding success, advertising effectiveness, online article sharing, retail sales, and movie box office returns.

The researchers suggest this approach might be particularly useful for markets involving “approaching positive outcomes” like entertainment and experiences, where positive emotional responses play a significant role in decision-making.

While the study doesn’t suggest that traditional market research should be abandoned, it offers a compelling complementary approach. By identifying which components of decision-making generalize most broadly across individuals, the findings also provide insights into human psychology and consumer behavior that extend beyond specific market applications.

As technology advances and neuroimaging becomes more accessible, these findings may open new avenues for understanding not just what consumers choose, but why—focusing on the emotional responses that appear to transcend demographic differences and connect us in our decision-making processes.