Real explains: "The economic, social, and political definitions of rationality are all 'consistency in choice behavior.' What it means to be rational is that an organism shows a consistent pattern of choices. There are definable rules that underlie that consistent pattern of observed choices."
Bumblebees, Real says, "are rational in that they have a consistent rule for choice, but they're not rational in a way that you might predict a priori." Apparently, the bumblebees' rule is to pursue only the short-term benefit. "We have shown consistently that bees seem to be using only short- term calculations, short-term rules, to make choices," Real says.
For example, Real has created situations in the lab in which one "flower" type offers a higher short-term energy gain, or nectar yield, and a second type offers the same energy gain over a longer time period. "In a choice situation between two artificial flower types, blue and yellow, when blue flowers and yellow flowers don't differ in the long-term rates of energy gain to the bee, but do differ in their short-term rates, bees consistently choose the flower with the highest short-term rate. The bees demonstrate a strong preference for short-term gain rather than long-term gain," Real says.
"Not only bees show this short-term pattern of choices," he adds. "Pigeons show it, starlings show it, and humans show it as well." A reliance on short-term calculation, Real suggests, is consistent with the idea of adaptiveness, an idea on which evolutionary theory is based.
"Given the environment that organisms live in," he has argued, "short-term processing of information may be quite adaptive. Environments are highly structured. In a large number of classes of phenomena, the best predictor of what is going to happen in the very near future is what has just occurred in the very recent past. That's probably why bees, as well as human beings, often use only short sequences of past events in their memory. Short-term sequences in memory are most predictive."
Relying on short-term experience, however, may not be as beneficial to humans in our current evolutionary stage as it is to bees in theirs. For example, says Real, "Humans stay out in the sun too long. In the short term, it has rewards, but in the long term, it's harmful." He adds, "That seems to be a problem with a lot of human behavior, that the scale over which humans make decisions is too short. I argue that we evolved from food-gatherers under conditions where food was probably spatially auto-correlated (concentrated in location) in the same way that the bees' food is spatially autocorrelated, and that long-term pooling, or retention, of information in a hunter-gatherer society is not very adaptive."
Most of us no longer live in such societies, however. And unlike flower patches, which have not changed substantially since bees' brains evolved, humans' environments have undergone radical change--much of it for the worse. Real suggests that as a species we need to begin thinking in longer terms.
But bumblebees still manage well by basing their decisions on short-term events, and Real analyzes these decisions, applying statistical formulas that at first glance may seem more suitable to a business classroom than a biology lab. According to Real, however, the decisions made by the bumblebees exemplify "the standard economic choice problems--the way someone would choose a stock portfolio." Characterizing the bees as investors looking for the best rate of return is one way to compare information processing in the bee with human cognition. Or, Real says, bees can be evaluated as gamblers, though bees and humans may not calculate odds in the same way.
A flower patch can be seen as a lottery, Real suggests, considering the fact that the mathematical models used to measure human lottery choices are identical to those he uses when he measures the bees' behavior. To describe humans' gambling decisions, he cites a set of classic experiments on probability assessment that were conducted by researchers Daniel Kahneman and Amos Tversky in the late 1970s and early 1980s. "Kahneman and Tversky have argued that human beings systematically misjudge probabilities by underrepresenting common events and overrepresenting rare events. For example, we think a plane crash will happen more frequently than it actually does," Real says. Bees, however, demonstrate a different assessment of probabilities. For example, when a highly rewarding flower type occurs rarely in an environment, it apparently does not enter into the bee's characterization of the environment, probably because it does not remain in memory. "Bees underrepresent rare events and overrepresent common events. They pay attention to the things that are common in their environment, and they ignore things that are relatively rare," Real says.
Actually, Real adds, it could be that humans do not overrepresent to the same degree that bees underrepresent. The Kahneman and Tversky experiments have been repeated in recent years, and the later experiments may indicate that humans can predict both rare and common events more accurately. The key to greater accuracy seems to be experience; humans are more likely to err when predicting the likelihood of events they have simply been told about than when predicting events they have learned about firsthand. "It is the limited ability of bees to recall long strings of past events that may lead to their misrepresenting probabilities," Real says.
Perhaps the bees' behavior is not surprising, if one hypothesizes that bees use only immediate experience to guide their behavior. "Humans can maintain a lot of information, whereas bees may have much more limited capacity for long- term calculation and long-term memory," Real says. "If bees' memories are extremely short, their probabilities are going to be biased in the direction of not representing very rare events and overrepresenting common events. They can't keep track of rare events when they occur because they are dropped from memory."
At this point, however, the bee's memory capabilities are still under investigation. "We don't know exactly what the long-term memory is in bees, or what the dynamic relationship is between short-term memory and long-term," Real says. "It's certainly clear that honeybees can remember things for 24 hours, or maybe even several days. But that's still a short time scale compared with human beings." And Real has chosen to focus not on honeybees, which have a complex social structure, strict division of labor, and a dance-language, but on bumblebees, which exhibit none of these features, precisely because of bumblebees' comparative simplicity. "If a scientist wants to look for the evolution of particular kinds of information processing machinery, or rules, in brains, that allow an organism to survive and reproduce efficiently in its environment, the bumblebee is an ideal organism. Its behavior is simple and goal directed," he says.
Just as bees and humans differ in decision-making patterns and memory capacity, their cognitive "machinery" differs vastly. Insect brains and are not totally unlike human brains, however. "The insect brain has some similarities with the human brain in terms of a common diffusive structure," Real says. "An input to the nervous system can often ascend into the insect brain and excite many different neurons simultaneously. Much cognitive processing in the human brain seems to follow this diffusive structure." Other comparisons are, at this point, difficult to make. For example, a bumblebee's method of locating the flowers it wants probably does not rely on the same kind of spatial memory that humans demonstrate. "Spatial memory is associated with the hippocampus in humans, and the bumblebee doesn't have a hippocampus," Real says. "The evidence for spatial memory in bees is in their behavior, the fact that they repeatedly come to specific locations."
To uncover more evidence, field research is needed, Real says. He plans to spend two summers gathering data about bumblebees' foraging behavior in natural settings. "We'll be trying to see if natural environments are characterized by having 'rugged energetic landscapes'--that is, landscapes with peaks and troughs: high nectar, low nectar, high energy, low energy." Real and his research associates will create formulas based on the patterns they observe in nature and feed them into a software program that mimics a bee's decision-making patterns based on what is known about honeybee neurophysiology.
These computer programs are being developed here at IU and in collaboration with neuroscientists at the Salk Institute in La Jolla, California. "We'll have computer bees with different computational algorithms associated with short-term versus long-term averaging and see what the best rules are for exploiting resources in these sorts of landscapes," Real says.
Computers are good at imitating mental processes, according to Real. "The computer has two components --the hardware and the software," he says. "The hardware (comparable to the brain's physical structure) determines how fast things can be done, how impulses are recorded. The software (comparable to the brain's 'program' for processing information) tells the computer what to do.
"The question is, how have the 'hardware' and the 'software' of brains coevolved, to efficiently process the information that's coming in through the senses in order to come up with a good solution to a problem? We've been trying to characterize what might be the 'software.' What are the algorithms that are implemented in the bee's brain to process the information efficiently? In answer to that question, we're discovering these short-term rules. The next question will be, how is the 'hardware' in the bee's brain structured so that it can implement those algorithms, those calculations?"
It is essential for Real to detect and fully describe such phenomena as the mechanisms of the bumblebee brain. This information adds to scientific understanding of the "cognitive ecology," which Real defined in a recently published article: "The research objective of cognitive ecology is to elucidate the underlying psychological and cognitive processes that enter into ecological decision-making, to determine the degree to which these mechanisms are the product of adaptive evolutionary change, and to ascertain the degree to which existing cognitive information-processing schemes constrain potential characterizations of the environment of the organism." Knowledge about the adaptive evolutionary change of bees contributes to knowledge about all adaptive evolutionary change, including our own.
See sidebar for A Rat's Tenacious Memory