brain and present at birth? Was it learned from induction because every time in the past, when any entity A was bigger than B and B was bigger than C, it was always the case that A was bigger than C as well? Or was it learned initially through language? Whether this ability is innate or learned, does it depend on some kind of silent internal language that mirrors and partially taps into the same neural machinery used for spoken language? Does language precede propositional logic, or vice versa? Or perhaps neither is necessary for the other, even though they mutually enrich each other.
These are intriguing theoretical questions, but can we translate them into experiments and find some answers? Doing so has proved to be notoriously difficult in the past, but I’ll propose what philosophers would call a thought experiment (although, unlike philosophers’ thought experiments, this one can actually be done). Imagine I show you three boxes of three different sizes on the floor and a desirable object dangling from a high ceiling. You will instantly stack the three boxes, with the largest one at the bottom and the smallest at the top, and then climb up to retrieve the reward. A chimp can also solve this problem but presumably requires physical trial-and-error exploration of the boxes (unless you pick an Einstein among chimps).
But now I modify the experiment: I put a colored luminous spot on each of the boxes—red (on the big box), blue (intermediate box), and green (small box)—and have the boxes lying separately on the floor. I bring you into the room for the first time and expose you to the boxes long enough for you to realize which box has which spot. Then I switch the room lights off so that only the luminous colored dots are visible. Finally, I bring a luminous reward into the dark room and dangle it from the ceiling.
If you have a normal brain you will, without hesitation, put the red-dotted box at the bottom, the blue-dotted box in the middle, and the green-dotted box on top, and then climb to the top of the pile to retrieve the dangling reward. (Let’s assume the boxes have handles sticking out that you use to pick them up with, and that the boxes have been made equal weight so that you can’t use tactile cues to distinguish them.) In other words, as a human being you can create arbitrary symbols (loosely analogous to words) and then juggle them entirely in your brain, doing a virtual-reality simulation to discover a solution. You could even do this if during the first phase you were shown only the red-and green-dotted boxes, and then separately shown the green-and blue-dotted boxes, followed finally in the test phase by seeing the red-and green-dotted boxes alone. (Assume that stacking even two boxes gives you better access to the reward.) Even though the relative sizes of the boxes were not currently visible during these three viewing stages, I bet you could now juggle the symbols entirely in your head to establish the transitivity using conditional (if-then) statements—“If red is bigger than blue and blue is bigger than green, then red must be bigger than green”—and then proceed to stack the green box on the red box in the dark to reach the reward. An ape would almost certainly fail at this task, which requires off-line (out of sight) manipulation of arbitrary signs, the basis of language.
But to what extent is language an actual requirement for conditional statements mentally processed off-line, especially in novel situations? Perhaps one could find out by carrying out the same experiment on a patient who has Wernicke’s aphasia. Given the claim that the patient can produce sentences like “If Blaka is bigger than Guli, then Lika tuk,” the question is whether she understands the transitivity implied in the sentence. If so, would she pass the three-boxes test we designed for chimps? Conversely, what about a patient with Broca’s aphasia, who purportedly has a broken syntax box? He no longer uses “ifs,” “buts,” and “thens” in his sentences and doesn’t comprehend these words when he hears or reads them. Would such a patient nevertheless be able to pass the three-boxes test, implying he doesn’t need the syntax module to understand and deploy the rules of deductive if-then inferences in a versatile manner? One could ask the same question of a number of other rules of logic as well. Without such experiments the interface between language and thought will forever remain a nebulous topic reserved for philosophers.
I have used the three-boxes idea to illustrate that one can, in principle, experimentally disentangle language and thought. But if the experiment proves impractical to carry out, one could conceivably confront the patient with cleverly designed video games that embody the same logic but do not require explicit verbal instructions. How good would the patient be at such games? And indeed, can the games themselves be used to slowly coax language comprehension back into action?
Another point to consider is that the ability to deploy transitivity in abstract logic may have evolved initially in a social context. Ape A sees ape B bullying and subduing ape C, who has on previous occasions successfully subdued A. Would A then spontaneously retreat from B, implying the ability to employ transitivity? (As a control, one would have to show that A doesn’t retreat from B if B is only seen subduing some other random ape C.)
The three-boxes test given to Wernicke’s aphasics might help us to disentangle the internal logic of our thought processes and the extent to which they interact with language. But there is also a curious emotional aspect to this syndrome that has received scant attention, namely, aphasics’ complete indifference—indeed, ignorance—of the fact that they are producing gibberish and their failure to register the expression of incomprehension on the faces of people they are talking to. Conversely, I once wandered into a clinic and started saying “Sawadee Khrap. Chua alai? Kin Krao la yang?” to an American patient and he smiled and nodded acknowledgment. Without his language comprehension module he couldn’t tell nonsense speech and normal speech apart, whether the speech emerged from his own mouth or from mine. My postdoctoral colleague Eric Altschuler and I have often toyed with the idea of introducing two Wernicke’s aphasics to each other. Would they talk incessantly to each other all day, and without getting bored? We joked about the possibility that Wernicke’s aphasics are
WE HAVE BEEN speculating on the evolution of language and thought, but still haven’t resolved it. (The three- boxes experiment or its video-game analog hasn’t been tried yet.) Nor have we considered the modularity of language itself: the distinction between semantics and syntax (including what we defined earlier in the chapter as recursive embedding, for example, “The girl who killed the cat that ate the rat started to sing”). Presently, the strongest evidence for the modularity of syntax comes from neurology, from the observation that patients with a damaged Wernicke’s area produce elaborate, grammatically correct sentences that are devoid of meaning. Conversely, in patients who have a damaged Broca’s area but an intact Wernicke’s area, like Dr. Hamdi, meaning is preserved, but there is no syntactic deep structure. If semantics (“thought”) and syntax were mediated by the same brain region or by diffuse neural networks, such an “uncoupling” or dissociation of the two functions couldn’t occur. This is the standard view presented by psycholinguists, but is it really true? The fact that the deep structure of language is deranged in Broca’s aphasia is beyond question, but does it follow that this brain region is specialized exclusively for key aspects of language such as recursion and hierarchical embedding? If I lop off your hand you can’t write, but your writing center is in the angular gyrus, not in your hand. To counter this argument psycholinguists usually point out that the converse of this syndrome occurs when Wernicke’s area is damaged: Deep structure underlying grammar is preserved but meaning is abolished.
My postdoctoral colleagues Paul McGeoch and David Brang and I decided to take a closer look. In an influential and brilliant paper written in 2001 in the journal
