Brain oscillations reveal dynamic shifts in creative thought during metaphor generation

When people generate creative metaphors for scientific ideas, their brains shift through distinct patterns of neural activity over time. A new study published in Psychophysiology used electroencephalography (EEG) and advanced modeling to track these brain dynamics and found that the most creative metaphors were associated with a specific sequence of brain states. Early in the process, higher activity in a brain state marked by alpha-band synchronization predicted greater metaphor novelty. Later on, this pattern flipped, with alpha-band desynchronization becoming more prominent. The findings offer new insight into how the brain supports creativity through time-varying electrical rhythms.

Researchers at the University of Arizona sought to better understand the neural mechanisms behind metaphor generation, a creative skill that plays an important role in how people understand complex concepts and communicate abstract ideas. While previous research has investigated creativity using tasks like alternative uses or story generation, less is known about how the brain supports the creation of metaphors—especially using tools that can capture the fast-changing dynamics of neural activity.

“Making metaphors is a powerful tool for people to learn and communicate difficult concepts,” explained study authors Vicky Tzuyin Lai (an associate professor and director of the Cognitive Neuroscience of Language Laboratory) and Yuhua Yu (a postdoctoral researcher in the Neuroscience of Emotion and Thought lab).

“Some metaphors are particularly effective because they are novel, clever and appropriate for the situation — in other words, they are creative. In this study, we aim to understand the brain mechanism of generating creative metaphors by leveraging some of the latest developments in neural analytical methods such as the latent states modeling.”

The researchers recorded EEG data from participants while they generated metaphors for science concepts, such as describing the cornea of the eye as a “windshield.” The team aimed to capture the oscillatory brain processes that underlie this form of verbal creativity. Instead of focusing on static measures of brain activity, they used a hidden Markov model to identify distinct “brain states” that participants cycled through while generating their metaphors. These states were defined by patterns of neural oscillations—rhythmic brain activity across different frequency bands.

Forty-three undergraduate students participated in the study. Each trial began with an audio description of a science concept, followed by a period in which participants generated a metaphor describing the concept’s function. After typing their metaphor, they rated it for novelty and aptness. Independent raters also evaluated the quality of each metaphor. During each trial, EEG data were collected from 32 scalp electrodes. The researchers used this data to identify recurring patterns of brain activity across trials and participants.

Using the hidden Markov model, the researchers extracted six distinct brain states, three of which were interpretable in terms of known oscillatory activity. One state was characterized by widespread alpha-band synchronization, typically associated with internal focus and suppression of irrelevant sensory input. Another showed widespread alpha desynchronization, which has been linked to increased arousal or attentional shifts. A third state featured gamma-band synchronization, often associated with bottom-up perceptual processing.

The researchers then examined how the amount of time participants spent in each state predicted the novelty and aptness of the metaphors they generated. Ratings were provided both by the participants themselves and by independent crowd raters. A clear pattern emerged: spending more time in the alpha synchronization state early in the metaphor generation process predicted higher novelty ratings. This finding was consistent across both self- and crowd-ratings. Participants also showed more frequent transitions into the alpha synchronization state when generating metaphors rated as more novel.

Interestingly, for metaphors that participants rated as more novel (but not those that received higher crowd ratings), more time spent in the alpha desynchronization state was also predictive. This effect occurred later in the generation process, suggesting that as people neared the point of formulating their metaphor, their brain activity shifted in a way that might reflect increased arousal or readiness to act.

The researchers also found that when participants generated metaphors they rated as highly novel, their brains showed more transitions between the alpha synchronization and desynchronization states. These findings suggest that creative metaphor generation may involve an interplay between focused internal attention early on and heightened readiness or arousal just before the creative output is finalized.

In contrast, the gamma-band state told a different story. In exploratory analyses, the researchers found that spending more time in the gamma state was negatively associated with metaphor novelty. This relationship held for both self- and crowd-ratings but only emerged when the gamma state was examined in isolation. This suggests that excessive gamma activity—possibly reflecting externally directed attention or sensory processing—may interfere with the internal thought processes needed for creative ideation.

Together, these results paint a picture of creativity as a dynamic, multi-stage process. At first, the brain may suppress distractions through alpha synchronization to allow for deep internal processing and idea exploration. Later, alpha desynchronization may mark a shift toward output and execution, as the idea solidifies. This temporal dance between different brain rhythms appears to support the emergence of novel, creative metaphors.

“Generating creative ideas to achieve a goal is a complex process,” Lai and Yu told PsyPost. “It evolves over time and sometimes involves paradoxical processes. To generate more creative metaphors, as we showed in the study, the brain first exhibits more synchronization in a type of neural oscillation known for inhibiting distractions. But later in the process, the oscillation flips the sign (i.e., more resynchronization), suggesting neural excitation immediately before reporting a creative metaphor.”

The study also highlights the benefits of using dynamic modeling approaches to capture the flow of brain activity over time. Traditional EEG analyses often average signals across trials or time windows, which can obscure the sequence of mental events during sustained thought. By using a hidden Markov model, the researchers were able to untangle these overlapping processes and better understand how the brain orchestrates complex cognitive tasks like metaphor generation.

“It is exciting to see the different involvement of the same neural process, known as the alpha-band oscillation, depending on what stage of generation a person is in,” the researchers said. “This finding consolidates previous theories. It is well known that the neural synchronization in alpha-band is critical for creativity. It has also been shown that the desynchronization in the same band marks excitation. We managed to consolidate the multifaceted roles of the same type of oscillation by focusing on the temporal dynamics in electrophysiological data that comes with high temporal resolution.”

Despite its strengths, the study has some limitations. While the results support a role for alpha oscillations in creativity, the findings related to gamma-band activity are more tentative and need to be replicated in future work. The crowd ratings of metaphor novelty also had lower reliability than ratings of aptness, which could have reduced sensitivity to detect some effects. Additionally, while the study focused on metaphor generation, it remains unclear whether similar brain dynamics would be observed for other types of creative comparison, such as literal analogies.

“We have conducted quantitative analysis of the brain data so far,” Lai and Yu explained. “The next step is to conduct qualitative analysis of the linguistic data, namely, the metaphors created by the study participants. By associating types of metaphors with patterns of brain oscillation, we hope to uncover finer grained details of verbal creativity in the brain.”

The study, “Hidden Brain States Reveal the Temporal Dynamics of Neural Oscillations During Metaphor Generation and Their Role in Verbal Creativity,” was authored by Yuhua Yu, Lindsay Krebs, Mark Beeman, and Vicky T. Lai.