June 20, 2012 | 11
Yogi Bear always claimed that he was smarter than the average bear, but the average bear appears to be smarter than once thought. Psychologists Jennifer Vonk of Oakland University and Michael J. Beran of Georgia State University have taken a testing methodology commonly used for primates and shown not only that the methodology can be more widely used, but also that bears can distinguish among differing numerosities.
Numerical cognition is perhaps the best understood of the core building blocks of the mind. Decades of research have provided evidence for the numerical abilities of gorillas, chimpanzees, rhesus, capuchin, and squirrel monkeys, lemurs, dolphins, elephants, birds, and fish. Pre-linguistic human infants share the same mental modules for representing and understanding numbers as those non-human animal species. Each of these species is able to precisely count sets of objects up to three, but after that, they can only approximate the number of items in a set. Even human adults living in cultures whose languages have not developed an explicit count list must rely on approximation rather than precision for quantities larger than three. For this reason, it is easier for infants and animals to distinguish thirty from sixty than it is to distinguish thirty from forty, since the 1:2 ratio (30:60) is smaller than the 3:4 ratio (30:40). As the ratios increase, the difference between the two sets becomes smaller, making it more difficult to discriminate between them without explicit counting.
Given that species as divergent as humans and mosquitofish represent number in the same ways, subject to the same (quantity-based and ratio-based) limits and constraints, it stands to reason that the ability to distinguish among two quantities is evolutionarily-ancient. That is, it is more likely that the ability emerged early in evolution than for the multiple cognitive systems to have developed identical limits and constraints multiple times over evolution.
Vonk and Beran argue, however, that there’s one problem with that argument. Most of the species that have been studied with respect to numerical cognition have been social: primates, cetaceans, and birds such as corvids and parrots. Given this, they reason, it is possible that the ability to estimate the number of items in a set may be related to the necessity for tracking members of the social group. A non-social animal – even a large-brained one such as the black bear – may therefore not possess this ability.
Alternatively, if numerical cognition is not related to sociality, then black bears should indeed be able to discriminate sets on the basis of numerosity. Further, their performance should decline as the ratio increases, just as for human infants, human adults from cultures without count lists, non-human primates, cetaceans, and birds.
One of the things that Venk and Beran were interested in was whether bears can discriminate sets of dots solely on the basis of their numerosity, rather than some correlated variable, such as surface area. Therefore, they created two different types of test trials. In congruent trials, the number of dots in each set was correlated with their surface area. In incongruent trials, the number of dots in each set was not correlated with their surface area.
Imagine that Yogi Bear spies two picnic blankets. On one blanket, there are three giant picnic baskets, and on the second there are two average-sized ones. If Yogi chooses the three giant baskets, it would be unclear whether he was choosing on the basis of number or volume, since they’re correlated. However, if he chooses three smaller baskets instead of two giant ones, then it is more reasonable to assume that the choice was based on number. Similarly, in this experiment, if the bears were only successful on congruent trials, then no conclusions could be drawn about their numerical abilities, per se.
But Vonk and Beran weren’t just interested in determining the numerical abilities of non-social carnivores. Because different methods are often used to assess the mental capabilities of different species, it is sometimes difficult to directly compare the results from different experiments. When it comes to working with primates and birds, researchers often use computerized stimuli displayed on touch screens. To signal a choice such as “this set has more objects,” the monkey or ape presses on the screen with a finger. Birds usually peck at it with their beaks. Vonk and Beran wanted to know if they could get bears to tap a screen with their noses. By making the experimental methodology as similar as possible across species, it becomes more reasonable to directly compare performance across species.
Three captive American black bear siblings (Ursus americanus) from the Mobile Zoo in Wilmer, Alabama, participated in the experiment. Researchers trained the bears to respond to a touch screen by pressing their noses against it. During training, Brutus was rewarded with a melody and with honey roasted peanuts, banana pellets, dried banana chips, yogurt-covered raisins, or wafer cookies after choosing the larger of the two sets, each of which contained between 1 and 10 dots of varying sizes. Dusty and Bella were rewarded for choosing the smaller of the two sets. When they chose the wrong option, the computer responded with a loud buzz.
The bears were most successful when the correct answer could be selected on the basis of both number and surface area, while they were significantly less successful when number was incongruent with surface area. This was particularly true for Bella and Dusty, who were rewarded for choosing the sets with fewer dots. It was apparently hard for them to overcome their desire to choose the set with the larger surface area.
However, even for incongruent trials, all three bears performed better than would have been predicted by random chance. The researchers reasoned that while surface area was perhaps more salient, the three bears were also able to represent numerical quantities. Surface area alone couldn’t have been responsible for their decisions. Number itself must have played at least some role in their performance.
Like all other species that have ever been tested, the bears’ performance was better when the ratio was small (and therefore, the numerical distance between the two sets was large) than when the ratio was large. They were better able to distinguish 2 from 10 (1:5 ratio) or 2 from 8 (1:4 ratio), for example, than 4 from 5 or 8 from 10 (both 4:5 ratio). The graph at right shows Brutus’s performance for the one of the conditions. As the ratio increases, his performance decreases towards random chance.
Together, these findings indicate that it may be easier for bears to choose larger amounts over smaller amounts. When considering picnic baskets, this strategy makes sense, but it isn’t clear why this preference would persist for abstract, two-dimensional dots on a computer screen, or when choosing the smaller amount is rewarded during training. Vonk and Beran say that choosing the greater amount or greater number of items may simply be more intuitive than choosing the smaller amount or number. This would be consistent with evidence that human children have an initial preference for “more.”
Several questions remain. What role might sociality have in shaping numerical cognition? How might the different needs of predator species compared with prey species impact the way that number is represented? One puzzling finding was that bears were less successful at identifying the set with more dots (for Brutus) or less dots (for Bella and Dusty) when the dots were moving instead of stationary. Why might this be?
Numerical cognition is a favorite topic among comparative psychologists, so these questions will no doubt be addressed in the coming years.
The main contribution of this research, however, is broader than the nuances of dot-counting. Often, researchers focus on species most closely related to humans (e.g. primates and domesticates such as dogs), or those easiest to test and house in a laboratory (such as fish and birds). By demonstrating that the minds of large carnivores such as bears can be probed using touch screens, other researchers may be encouraged to expand their research programs to include bears and other under-studied species.
“It is exciting,” Vonk and Beran write, “to consider that such divergent species can be tested in the same way to promote a fuller picture of comparative cognition and the diverse forces giving rise to both similar and distinct traits.” In that sense, their paper is a comparative call-to-arms. In order to more completely understand the way that minds are made, a broader approach must be considered, which begins by studying species that occupy social and physical ecological niches that are different from those that are most familiar.
Vonk, J., & Beran, M. J. (2012). Bears ‘count’ too: quantity estimation and comparison in black bears, Ursus americanus. Animal Behaviour DOI: 10.1016/j.anbehav.2012.05.001
Secrets of the Universe: Past, Present, FutureX