December 26, 2012
They called him “Diogenes the Cynic,” because “cynic” meant “dog-like,” and he had a habit of basking naked on the lawn while his fellow philosophers talked on the porch. While they debated the mysteries of the cosmos, Diogenes preferred to soak up some rays – some have called him the Jimmy Buffett of ancient Greece.
Anyway, one morning, the great philosopher Plato had a stroke of insight. He caught everyone’s attention, gathered a crowd around him, and announced his deduction: “Man is defined as a hairless, featherless, two-legged animal!” Whereupon Diogenes abruptly leaped up from the lawn, dashed off to the marketplace, and burst back onto the porch carrying a plucked chicken – which he held aloft and shouted, “Behold: I give you… Man!”
I’m sure Plato was less than thrilled at this stunt, but the story reminds us that these early philosophers were still hammering out the most basic tenets of the science we now know as taxonomy: The grouping of objects from the world into abstract categories. This technique of chopping up reality wasn’t invented in ancient Greece, though. In fact, as a recent study shows, it’s fundamental to the way our brains work.
Chunks of reality
At the most basic level, we don’t really perceive separate objects at all – we perceive our nervous systems’ responses to a boundless flow of electromagnetic waves and biochemical reactions. Our brains slot certain neural response patterns into sensory pathways we call “sight,” “smell” and so on – but abilities like synesthesia and echolocation show that even the boundaries between our senses can be blurry.
Still, our brains are talented at picking out certain chunks of sensory experience and associating those chunks with other stimuli. For instance, if you hear purring and feel fur rubbing against your leg, your brain knows to associate that sound and feeling with the fluffy four-legged object you see at your feet – and to group that whole multisensory chunk under the heading of “cat.”
What’s more, years of cat experience have taught you that it makes no sense to think of a cat as if it were a piece of furniture, or a truck, or a weather balloon. In other words, an encounter with a cat carries a particular set of meanings for you – and those meanings determine which areas of your brain will perk up in the presence of a feline.
But where’s the category “cat” in the brain? And where’s it situated in relation to, say, “dog” or “giraffe” …or just “mammal?” A team of neuroscientists led by Alexander Huth at UC Berkeley’s Gallant lab decided they’d answer these questions in the most thorough way possible: By capturing brain responses to every kind of object they could dig up.
Chunks in the brain
Those Gallant lab folks are no slouches – you might remember them as the lab that constructed “mind videos” of entire scenes from neural activity in the visual cortex. This time, though, the lab’s ambitions were even broader.
A research team led by Alex Huth showed volunteers hours of video footage of thousands of everyday objects and scenes – from cats and birds to cars and thunderstorms – as the subjects sat in an fMRI scanner. Then the researchers matched up the volunteers’ brain activity not only to each object they saw, but also to a whole tree of nested object categories: A taxonomy of the brain’s taxonomy. A vision of a “continuous semantic space,” where thousands of objects and actions are represented in terms of others.
Huth’s team collected volunteers’ reactions to more than 1,300 objects and categories, and arranged these brain responses not only into a tree of object and action categories, but into a map of response gradients across the whole surface of the brain.
And as you can see from the color gradients in that tree diagram to the right (which is also available as an interactive online app), the relationships among our brains’ categories are multidimensional. Objects may be more or less “animal-like,” more or less “man-made,” and so on – and in fact, the researchers say they expect to find more subtle response dimensions that gauge an object’s size and speed.
Association and meaning
All this talk of “dimensions of association” points back to a far more profound idea about how our brains work: We understand the meaning of an object in terms of the meanings of other objects – other chunks of reality to which our brains have assigned certain characteristics. In the brain’s taxonomy, there are no discrete entries or “files” – just associations that are more strongly or more weakly correlated with other associations.
And that idea itself raises deeper quandaries: If associations define what an object or action “is,” as some neuroscientists have argued, then why does the concept of meaning – semantic representation – need to enter the picture at all? Instead of being a special type of mental function, might “meaning” itself simply be another word for “association?”
The answer to that question won’t be a simple one to find, at least for the foreseeable future. “I don’t think it’s possible to make a conclusive claim about that from fMRI data,” says Jack Gallant, the lab’s director; “and anyone who tells you otherwise is mistaken.”
A single three-dimensional pixel – an fMRI voxel – represents the activity of around one million neurons, Gallant explains; and at that resolution, it’s impossible to say what exactly the neural activity is encoding. Meaning could depend on association, association might depend on semantic coding, or the relationship between the two might be more nuanced than we can conceive right now.
Whatever that relationship turns out to be, the implication remains: In our brains, meaning and association go hand-in-hand. In the brain, even our most abstract concepts depend on our own real-world experiences. That’s an idea that’s infuriated Plato and his followers far more than Diogenes’ plucked chicken – but as Diogenes demonstrated on that long-ago morning, real-world evidence trumps speculation in the end.