The paper, by the way, was by Andras Liker and Veronica Bokony. It appeared in the Proceedings of the National Academy of Sciences, 106:7893-7898, entitled, "Larger groups are more successful in innovative problem solving in house sparrows".
After collection, the birds were banded (on their legs) so the investigators would know which birds came from what area, and the exact date they were collected. They were brought to an aviary, and kept all together, fed seeds from open bowls, as much as they liked (ad libidum). The birds were kept in the aviary for at least a month before the experiment began, and then at the time of the experiment (one experiment per week), they were separated randomly into different groups of 6 and 2, some of which were from urban setting, some from the country, some were mixed gender, some were all males, some all females.
Each experiment lasted for 3 days. On the first day the open bowls of food were taken away, and in their place was a transparent plexiglass feeder with 16 holes on top, 3.5 cm in diameter, from which the birds could access food. At night the feeder was removed, food was replenished and the feeder was brought back the next morning.
Importantly, mounted on the feeder were 16 transparent caps with little black knobs. These would be placed over the holes on the feeder on the morning of the third day. Why is that important? Because this way by the third day the birds are familiar with the looks of the caps on the feeder, but until the 3rd day they didn’t cover anything; they were just there. So the first day represented basal, or control, conditions. The birds could access food, they saw the caps, and there was no problem-solving. On day one the birds became acclimated to the feeder and to the transparent caps that were attached but didn’t cover anything.
On the second day, the investigators attached a weird looking item, which they describe in detail, poking out from the feeder. This was meant to scare the birds a little bit, until the birds discovered that whatever it was, it was harmless. They called this a test for neophobia. They did this to see whether being in a group resulted in decreased fear, measured by taking less time than small groups to approach the feeder. They also measured how much time birds in large and small groups scanned the environment, as for predators. In both cases they found no per-capita difference between large and small groups. This ruled out the possibility that large groups were more effective than small groups because they spent less time in fear and worry, and could therefore spend more time attacking the cap.
And on the third day came the key experiment. Each of the 16 holes was covered with one of the transparent caps. When the birds came to the feeder they could not access food until a cap was removed. Would larger groups be more effective in cap-removal? And if so, why?
Why did they do this experiment? What led them to this design? What were they expecting to find? According to their Introduction, they were interested in the important issue in ecology of the benefits and costs of group living vs solitary living. In groups, there is increased competitiveness, but that disadvantage could be offset by certain advantages, such as, a) individuals in a group can learn from one another when, where, what, and how to forage, b) they can exploit food resources discovered by others, c) individual risk of predation decreases in groups, d) time spent in scanning for predation can be decreased, allowing for an increase in foraging time and energy, and e) for hunters, there is higher success.
Groups could also adapt to novel situations more readily, either because they could cooperate, or without cooperation, could benefit from one another’s skills and experience, and possibly copy and learn from it. While this sounds good, the authors note that, “the effect of group size has not been experimentally studied in any task that required the invention of novel animal innovations, and may be important in adapting to novel environments.”
So that sets the stage for this study. On the face if it, it seems like a good study, one that speaks to a major problem in ecology. But, as I suggested earlier, I think the results may be much more significant. I think that what they have shown is that social interactions elicit greater individual problem-solving ability. And if that is the case, if cognitive skills can be increased because of a group setting, then in principle those skills could be directed at the organization of the group itself, which in turn could facilitate further cognitive successes; this positive feedback loop could accelerate, and could possibly account for the evolution of consciousness. I want to continue to explore this by looking at the results of the experiment, and examining how the investigators interpret them.
How were the observations made and what were the observations? Observations were made in two ways. First, there were video cameras recording everything that went on, and then the videos were studied afterwards. And for an hour and a half, the investigator made direct observations behind a one-way glass. After an hour and a half, the experiment was concluded.
The birds found that they could open a well in one of two different ways. They could push at the knob until the cap fell off, or they could peck at the edge of the cap until it loosened and fell off.
Most of the action occurred within the first 30 minutes. By that time wells were opened, and regardless of which bird opened a well, all the birds of the group were feeding. So ecologically it is beneficial for birds to feed in a group, and the larger the better, at least up to 6 in the group. While interesting, that is not what concerns me here. I am more interested in the problem-solving ability.
Here is a key sentence, perhaps THE key sentence, in their Results. “Large groups were significantly more successful in all aspects of problem-solving: they opened 4 times more wells in total, and opened the first well 11 times sooner than small groups.”
Now when you look at that sentence, you can see right away that there are two different types of abilities being described. In the first, the total number of wells opened, both the skill of the original opening a well as well the ability of the other birds to copy that skill for subsequent openings, are described. Once one bird opens a well the other birds see how to do it and can then open more wells. It follows that since there are more birds in the larger group, it is more likely that the larger group will open more wells. One might expect the larger groups to open three times as many well as the smaller groups. The actual number was 4 times as many, not too far off.
But opening the first well -- that can’t depend on copying someone else. That depends on a new application of a skill. One bird, and only one bird, could be the first to open a well. And their results show that larger groups were 11 times more successful than smaller groups. Not 3 times more successful, but 11 times.
Now consider this comment from the Discussion. “An interesting aspect of our results is that large groups were disproportionately more successful in the problem-solving task than small groups. If all birds are trying to get food independently of each other, thrice as many birds should mean thrice as many attempts and assuming random success, thrice as many problem-solving events. According to this expectation, large groups spent proportionately more time on the feeder, and made proportionately more attempts to problem solve than smaller groups. Although this may partially explain the higher success of the large groups it cannot account for their >10-fold faster problem solving… The latter result suggest that the effectiveness of all individuals was increased in the large groups and/or there were a greater number of effective individuals in large groups.” [italics mine].
I italicized this to show that the authors speculated on exactly the same interpretation I offered, namely that there is something about large groups that increases the cognitive effectiveness of the individual in the group. They say “all” the individuals, but I don’t think that was demonstrated. Rather, the cognitive effectiveness of at least ONE individual in the group was increased.
The last part of their last sentence is baffling to me: “or there were a greater number of effective individuals in large groups”. This cannot be. Probabilistically, there are 3 times the numbers of effective individuals in larger groups. Thrice as many effective birds in large groups should mean thrice as many problem-solving events, but they found 11 times the number of well-openings in larger groups. Their alternative “explanation” does not explain their results.
The authors present two possible explanations for their data; one that is really interesting, and the other that is patently false.
Nonetheless, they conclude the paper (and this is the conclusion picked up by the brief note in Science that I discussed last week) by saying “A likely reason for the disproportionate faster and more problem-solving by larger groups is their greater chance to contain skilled individuals, whose quick innovations might then further enhance the group’s success via social facilitation and social learning”.
Isn't that fascinating! They dropped altogether the first, more interesting possible conclusion they spoke of, that individual effectiveness increases in larger groups. Instead, they repeat the second explanation, that larger groups have a greater chance to contain skilled individuals. As mentioned above, they cannot contain more than three times the number of effective individuals, and therefore that cannot account for the 11-fold increase in the speed at which larger groups opened the first cap. And while their assertion that group success is increased via social facilitation and social learning clearly applies to the issue of more problem solving, it clearly CANNOT apply to the issue of faster problem solving.
I wonder if the authors were afraid of their first possible explanation. They probably viewed it as a formal possibility, but one that could not be readily understood, and therefore dropped it. They were probably at a loss when considering how “groupness” could possibly enhance the cognitive abilities of effective individuals in the group. But that is the exciting behavioral phenomenon that this study revealed, and it is exactly the question that we should now be asking.
This study showed that when it comes to problem-solving, 6 is clearly greater than 3x2.
I fear that I may be missing something here, so I am going to write to the authors for their reaction to this analysis.
Stay tuned.
again, very good clear writing and marvelously conjecturing sceince. Deep questions put in a grounded context--Hungary map and all. I am eager for more bio buzz!
ReplyDeleteV