The structure of the language shapes colors addition links even for people born blind, reveals the study

Humans are known to do mental associations between various stimuli and real world concepts, including colors. For example, red and orange are generally associated with words such as “hot” or “hot” blue with “fresh” or “cold” and white with “clean”.

Interestingly, some studies in psychology have shown that even if some of these associations arise from the direct experience of people to see the colors around the world around them, many people born blind always make associations of similar colors. The processes underlying the formation of associations between colors and specific adjectives have not yet been fully elucidated.

Researchers from the University of Wisconsin-Madison recently carried out a study to deepen how language contributes to the way we learn color, using mathematical and computer tools, including the GPT-4 Open AI (LLM) model. Their results, published in Communications psychologySuggest that color administration associations are rooted in the structure of language itself and are therefore not only learned by experience.

“Some colors are strongly associated with certain adjectives (for example, red is hot, blue is cold),” wrote Qiawen Liu, Jeroen Van Paidon and Gary Lupyan in their article.

“Some of these associations are based on visual experiences such as seeing brilliant red embers. Surprisingly, despite no visual experience, many congenitually blind people present very similar color associations, which are probably learned by language. We show that these associations are indeed anchored in the statistical structure of language.”

To explore the contribution of language to learning colored addition associations, Liu, Van Paridon and Lupyan used word interests. These are mathematical models that represent models in the way words are used in a set of written texts.

Using these models, the researchers have mapped colored administration associations in a set of data containing texts written in English. Subsequently, they compared the predictions made by their models to the associations made by blind and seen English -speaking individuals.

“We apply a projection method to the interests of the words formed on spoken and written language corpora to identify the associations of addition of colors such that they are represented in English,” wrote the researchers. “These projections were predictive of the associations of addition of colors reported by blind and seen English speakers.”

The researchers evaluated the extent to which the ancients of words derived from the text of fiction have captured associations of administration of colors generally made by blind and seen individuals. They also compared the predictions based on these interests to those manufactured by the Grand Language Model of Openai (LLM) GPT-4, which feeds the famous Chatgpt conversational platform.

They notably found that the predictions made by integrations of words derived from fictional texts were more closely aligned with the associations made by humans, outperforming the GPT-4 model. Liu, Van Paridon and Lupyan could also identify the sentences in the texts they analyzed which seemed to contribute the most to the learning of color administration associations.

“By increasing the formation corpora in various ways, we discover the most responsible types of sentences for the transmission of colored addition associations with the models,” said Liu, Van Paridon and Lupyan.

“We note that the models of incorporation of words learn of these associations from indirect coccecree (second -rate), and that when invited, people are able to identify some of the most informative words to associate colors with specific adjectives.”

Overall, the results collected by this team of researchers suggest that language plays a key role in the way humans learn to connect specific qualities to colors, regardless of differences in their perceptions and experiences. More specifically, this suggests that these associations are often learned via second -rate coccecuries, which are indirect connections between words.

For example, rather than learning to associate “red” with “hot” after having encountered many sentences which include these two words (for example, “the stove is red”), people could bind these terms because they have often found the word “red” in the same sentence as other words related to heat (for example, “fire” and “flame”). These third words thus dumb the relationship between an adjective and a color.

The recent study offers new precious information on the processes underlying the learning of associations of people who have had different life experiences. In the future, he could inspire other studies that compare real world observations to predictions made by computer and mathematical models to better understand psychological processes.

© 2025 Science X Network