I asked Brewer if that meant what I thought it did: had he actually asked people to compete against rats in a maze, almost literally running in a rat race, and found that the people did no better than the rats?
He had. I begged him, then, to please, please let me come out and run the maze, too. He was willing when we spoke, but later his collaborators turned out to be reluctant. They were concerned, I think, that the media might sensationalize this idea of a “man vs. mouse” intelligence competition.
“It is kind of eyebrow raising, I guess,” Brewer told me. “But to step out on a limb and then say that because the rats learn this maze as well as humans do, that we have similar cognitive abilities, that’s probably not a good idea. We just happened to notice a peculiar similarity between how well our volunteers learned it and how well the rats did. To really answer whether different species have similar types of cognitive repertoires would take much more work. The bugaboo is definitely language and symbolic thought. We can always ask people to give verbal reports of the strategies they use to complete a task, but you can’t do that with animals. The thing that’s tough about animal cognition, you’re studying a thing that you can’t detect except through behavior.”
Still, Brewer was happy to explain how the study was designed.
Radial mazes for rats are generally just a few feet wide, with tunnels extending from a center spoke like the arms of a Ferris wheel. Food is placed at the end of each arm, and the rats are tested to see if they can remember which of the arms they have already gone down, without repeating. For his study, Brewer added a twist to his eleven-arm radial maze to make it especially challenging: the rats would find no food pellets if they ran down each arm in order, the equivalent of going along a clock face from one o’clock to two o’clock to three o’clock. Nor would they find any if they went down every other arm, as if going from one o’clock to three o’clock to five o’clock. Rather, they had to go down every third arm of the maze, the equivalent of going from one o’clock to four o’clock to seven o’clock. Hard as that was for a rat to learn, the real trick was to remember which ones they had already gone down after the first circuit.
To replicate the same test in people, Brewer and his colleagues constructed the walls of an eleven-arm radial maze on a basketball court, out of plastic tarps. Center court was where the eleven arms of the maze met in the middle. At the end of each arm was money.
“But it wasn’t money they were going to keep,” Brewer said. “We’ve now collected data on 150 participants. Animals need a reward to motivate them, but humans usually find motivation in just trying to succeed at a task. That’s a big difference between rats and humans.”
Yes, that’s one of them. But what was it like for the participants who tried to complete the maze?
“When you’re going through that maze, man, you’re good for about seven or eight arms,” Brewer said. “And then you’re, like—what? Where am I? You just lose it. As you start coming around the second time, it’s just very hard to keep track of eleven arms when you can’t do it in a simple order.”
Although Brewer was reluctant to infer from his study any conclusions about the intelligence of mice, a neuroscientist at Rutgers University in New Jersey had recently come out of the closet on the question.
“I once was reluctant to use the term ‘intelligence’ in mice, because it’s such a loaded term,” said Louis D. Matzel when we met at his office in the psychology building on the university’s Piscataway campus. “The first grant proposal I submitted to NIH, in 1992, used the word. The reviewer said there is no place in biology for studying intelligence because everybody knows that intelligence is just a social construct. That gives you a sense of why some animal scientists are so reluctant to discuss it. I even had a dean here at Rutgers who once told me I shouldn’t discuss intelligence because it’s such a difficult topic. But she was an idiot and got fired soon after, so I felt vindicated.”
I was beginning to like this guy. Matzel is a wiry middle-aged character with graying brown hair and a trim mustache. He wears Converse high-top sneakers, smokes cigarettes outside his building, and likes to climb mountains in the winter with his teenaged son. Formerly married to fellow Rutgers psychologist Tracey Shors (the one described in chapter 5 who made the quip at the “Cognitive Enhancers” meeting about an exploding head), Matzel had Christmas lights hung over his office windows— in September. “About five years ago I got too lazy to take them down,” he said. Lying atop a bookshelf was a CD called Until We’re Dead by a band I’d never heard of, Star Fucking Hipsters. On the cork wall near his desk was a photograph of the late Sid Vicious, bassist of the Sex Pistols, and his late girlfriend, Nancy Spungen, whom Sid had been suspected of murdering. Beside it was a photograph of Ian Curtis, the late singer of the British neo-punk band Joy Division. “You don’t know Joy Division?” he said. “Look them up on YouTube. There’s an excellent documentary that came out about them a few years ago.”
Not all his photographs were of dead punk rockers. “Right behind you is a picture of a snail,” he said. “Hermissenda is a pretty little sea snail. But that was a terrible field I was in. I had some ideas about mice around the year 2000, so I stopped working with snails and never looked back.”
Matzel’s latest studies of mice have offered perhaps the most astonishing replication imaginable of Jaeggi and Buschkuehl’s findings. Before he could conduct the studies, though, Matzel first had to design a test from scratch that would accurately measure working memory in mice. Not only did no such test exist, no proof really existed that mice have a working memory. Most researchers, like Matzel, assumed that a mouse needed to have a working memory to keep track of things in its environment. But nobody had ever really proved it.
To do so, Matzel used an apparatus so obscure, it had been described only once in a 1981 paper: the dual maze, consisting of two eight-armed radial mazes placed side by side in a small room whose walls are decorated with bold designs. The idea was to allow the mice to explore a few arms of one maze, then place them in the second to explore a few arms of that one, then put them back in the first — and see if they could remember which arms they had already explored in that first one, so that they wouldn’t repeat. Like sailors navigating the ocean by the stars, the mice could look up beyond the confines of whichever maze they were currently in to see the walls of the room and thereby orient themselves to the room as a whole— and, thereby, to the other maze. In a small laboratory across from his office, Matzel decorated the western wall with a large black S and a dangling string of tiny lights. On the eastern wall he hung a large black plus sign and another string of lights, but with fatter bulbs. He placed a poster of crudely drawn star shapes on the southern wall and an absurd drawing of cartoon characters from an Adult Swim cartoon on the northern wall.
“The thing is,” Matzel said, “if I ask the mouse to work on only one maze, he’s really good at it. A well-trained animal will typically make no errors. He’ll navigate around and get his eight pieces of food. With two mazes, it’s really hard. At first they all make many errors. But here’s what we found: over days of trying it again and again, they get good at it. A smart mouse will eventually get really good at it and make no errors. This is a task where some rodents are at least as good as humans. So they do have a working memory.”
I asked him what he meant by a “smart” mouse.
“We test intelligence in the mouse on a battery of six learning tests, each one different from the other. Occasionally, we get an animal who turns out to be the best on all six tests compared to fifty other mice. That’s a really smart mouse, we would say. We teach them to avoid a shock or avoid a bright light. Navigate through a dry maze. Navigate in a water maze. We also have a reasoning task.”
The reasoning task requires the mice to make an inference by exclusion. “I show an animal a star symbol,” he said, “and it learns to walk over to an object shaped like a circle; underneath is a treat. So the animal learns that when it sees the star, if it goes to the circle, it gets a piece of food. Star means circle. Then I train it on a square and a triangle. If you see a square, you’ll get food under the triangle. So square means ‘go to triangle.’ And then one day I show the animal a symbol it’s never seen before, let’s say a crescent, and out in the field of objects it can choose from, it might have a triangle, might have the circle, and a novel object it’s never seen before. It looks at the triangle and the circle and thinks, ‘It can’t be under those, so it must be under the novel object.’ It infers by exclusion that the food must be under the novel object. This is the amazing thing— mice are decent at doing that. This kind of reasoning task is considered a quintessential example of humans’ ability to reason. And yet mice do it. I’m so amazed they can do it. Because I don’t believe my dog can do it. My dog just seems really stupid. For years I’d let my dog out in the yard on a rope, and he’d always wrap himself around a tree.”
Using the reasoning task, Matzel demonstrated that, just as in humans, a general intelligence factor can be discerned in mice: those who are better at the reasoning tasks tend to be faster at learning the other tasks. Likewise, those who do better at the working-memory task as measured by the dual maze also tend to do better on the reasoning and learning tasks. But where things get really interesting is in Matzel’s amazing mouse version of Jaeggi and Buschkuehl’s training studies: in 2010, he reported that the animals whose working memory he trained by having them practice on the dual maze actually got smarter on tests of general cognitive abilities. Finally, and most significantly of all, mice who trained on the dual- maze task when they were younger showed less age-related loss of attention and learning abilities by the time they had reached the mouse equivalent of old age. Matzel and colleagues concluded in that study: “These results suggest that general impairments of learning, attention, and cognitive flexibility may be mitigated by a cognitive exercise regimen that requires chronic attentional engagement.” Or, as he told me, “It was our intention to manipulate working memory in mice and see if that manipulation had a direct effect on their intelligence. In fact, Jaeggi’s work demonstrates the same thing we found.”
If working- memory training increases intelligence in mice, imagine what it might have done for Sid Vicious.
President Obama says he will veto proposed bipartisan legislation to impose new sanctions on Iran so long as diplomatic negotiations over a nuclear deal remain underway.
PHOTO: Iranian Foreign Minister Mohammad Javad Zarif shakes hands with U.S. Secretary of State John Kerry as former EU foreign policy chief Catherine Ashton looks prior to closed-door nuclear talks with Iran in Vienna, Austria, Nov. 20, 2014; While the gathering was closed to reporters, Menendez was said to have taken 
Remember the Fox television show Are You Smarter Than a 5th Grader? In November 2011, results of the ultimate extreme version of that kind of contest were reported in Seattle at the annual meeting of the Psychonomic Society. Gene Brewer, the Arizona State researcher who collaborated on the Facebook study mentioned above, was the first author of a paper entitled “Working Memory in Rats and Humans.” According to the abstract, “The quintessential instrument used by animal researchers for measuring working memory in rodents is the radial-arm maze. We constructed an 11-arm human version of the radial-arm maze and assessed individual differences in maze ability, general-fluid intelligence, and working memory capacity.” A key phrase in the abstract’s conclusion caught my attention: “Human behavior in the maze paralleled that found in rodents.”
