When you think about a cup of coffee, what exactly are you thinking about? What sort of representations are you accessing in your memory? Tactile? Olfactory? The phrase “cup of coffee”? What makes the concept of a cup of coffee different from other concepts? These are all very difficult questions, and they get at one of the core issues of psycholinguistics: the relationship between language and thought.

This is the first post in a three-part series on Gary Lupyan’s label-feedback hypothesis.

I read and write about this topic a lot, but I recently came across a new perspective on this issue that I found to be very intriguing and compelling, if a bit hard to wrap my head around. This is Gary Lupyan’s “label-feedback hypothesis”, which states that verbal labels are co-activated when we activate conceptual representations — which means that if you think about the smell of a cup of coffee, you are activating the linguistic representation of the word “coffee” as well as the conceptual representation. I will come to why this is important in a moment.

Back to the label-feedback hypothesis and how it works: Let us talk about categorization for a second. When you see something — say a motor vehicle — you automatically categorize it. Is it a car? A van? A motorcycle? A bus? A scooter? A pickup? We know that this happens very quickly and without any conscious thought, but exactly how it happens is not well-understood. One of the common models is an exemplar model, in which objects are essentially defined by diagnostic features — for example, one diagnostic feature of a motorcycle is that it has two wheels. Maybe a sliding door is a diagnostic feature of a van.

Anyway, it is easiest to think about categorization as comparing what you see to a (real or imaginary) object that contains the diagnostic features. For example, one of the diagnostic features of cow could be that it has spots. If this is the case, when you see an animal, and it has spots, you will be more likely to classify it as a cow. If it neighs, you’ll be more likely to classify it as a horse, as this is a diagnostic feature of a horse.

In Lupyan’s model, when you activate a conceptual representation, the linguistic label of that category is also activated, and this highlights the diagnostic features of categories, essentially making categories seem more different than each other.

This is a lot of heavy cognitive talk, but what it comes down to is that it is the linguistic label that helps us categorize. The evidence for this comes from patients who have undergone brain trauma and have decreased linguistic abilities. Interestingly, people who are engaging in linguistic tasks (like repeating a string of nonsense syllables) behave quite similarly to aphasic patients in that they have difficulty in categorizing objects when they have to abstract qualities over sets of objects.

For example, when shown three pictures, two of red things and one of a blue thing, they have trouble grouping the red things together and saying that the blue one is the odd one out. However, if the objects are grouped by a theme — for example, a car, a truck, and a flower, the participants have much less trouble identifying the odd one.

The fact that linguistic interference disrupts the putative effect of language on non-linguistic thought can be seen as a challenge to traditional formulations of linguistic relativity, but Lupyan’s label-feedback hypothesis turns this challenge on its head and uses it as support for this particular instantiation of linguistic relativity.

More on how this works and what it means in the next post!

References

Lupyan G (2012). Linguistically modulated perception and cognition: the label-feedback hypothesis. Frontiers in psychology, 3 PMID: 22408629

Image via Elena Elisseeva / Shutterstock.