Â鶹ÒùÔº

Humans are known to make mental associations between various real-world stimuli and concepts, including colors. For example, red and orange are typically associated with words such as "hot" or "warm," blue with "cool" or "cold," and white with "clean."

Interestingly, some past psychology studies have shown that even if some of these associations arise from people's direct experience of seeing colors in the world around them, many people who were born blind still make similar color-adjective associations. The processes underpinning the formation of associations between colors and specific adjectives have not yet been fully elucidated.

Researchers at the University of Wisconsin-Madison recently carried out a study to further investigate how language contributes to how we learn about color, using mathematical and computational tools, including Open AI's GPT-4 large language model (LLM). Their findings, in Communications Psychology, suggest that color-adjective associations are rooted in the structure of language itself and are thus not only learned through experience.

"Certain colors are strongly associated with certain adjectives (e.g., red is hot, blue is cold)," wrote Qiawen Liu, Jeroen van Paridon, and Gary Lupyan in their paper.

"Some of these associations are grounded in visual experiences such as seeing glowing red embers. Surprisingly, despite having no visual experience, many congenitally blind people show very similar color associations, which are likely learned through language. We show that these associations are indeed embedded in the statistical structure of language."

To explore the contribution of language to the learning of color-adjective associations, Liu, van Paridon and Lupyan used word embeddings. These are mathematical models that represent patterns in how words are used within a set of written texts.

Using these models, the researchers mapped the color-adjective associations in a dataset containing texts written in English. Subsequently, they compared the predictions made by their models to associations made by both blind and sighted English-speaking individuals.

"We apply a projection method to word embeddings trained on corpora of spoken and written language to identify color-adjective associations as they are represented in English," wrote the researchers. "These projections were predictive of color-adjective associations reported by blind and sighted English speakers."

The researchers evaluated the extent to which word embeddings derived from text of fiction captured color-adjective associations typically made by both blind and sighted individuals. They also compared predictions based on these embeddings to those made by OpenAI's large language model (LLM) GPT-4, which powers the renowned conversational platform ChatGPT.

Notably, they found that the predictions made by word embeddings derived from fiction texts were more closely aligned to the associations made by humans, outperforming the GPT-4 model. Liu, van Paridon and Lupyan could also identify sentences within the texts they analyzed that appeared to contribute most to the learning of color-adjective associations.

"By augmenting the training corpora in various ways, we discover the types of sentences most responsible for conveying the color-adjective associations to the models," explained Liu, van Paridon and Lupyan.

"We find that word embedding models learn these associations from indirect (second-order) co-occurrences, and that when prompted, people are able to identify some of the words that are most informative for associating colors with specific adjectives."

Overall, the findings gathered by this team of researchers suggest that language plays a key role in how humans learn to connect specific qualities to colors, irrespective of differences in their perceptions and experiences. Specifically, it suggests that these associations are often learned via second order co-occurrences, which are indirect connections between words.

For example, rather than learning to associate "red" with "hot" after encountering many sentences that include both these words (e.g., "the stove is red"), people could link these terms because they often found the word "red" in the same sentence as other words that are related to heat (e.g., "fire" and "flame"). These third words thus mediate the relationship between an adjective and a color.

The recent study offers new valuable insight into the processes underlying the learning of associations by people who had different life experiences. In the future, it could inspire further studies that compare real-world observations to predictions made by computational and mathematical models to understand psychological processes better.

© 2025 Science X Network