What Social Insects Tell Us About Social Networks
You wouldn't think the age-old roommate battle over who's doing the dishes would have much to do with ant and bee research, but you're wrong! Ha! We win again!
That's because we spent last night listening to ASU's Jennifer Fewell and UW's Sean O'Donnell at the UW's Edwards Psychology Lecture series, and you were doing whatever you were doing and not learning about insects. One thing Fewell has unearthed in her research on ants is that they come with different response thresholds for tasks, and this ends up creating "specialists" in a colony. How dat? We will explicate.
Fewell studied putting ants together who were big excavators (that's a big ant job, digging) with some who were more, Hey, I'm gonna hang out here and...uh...get some stuff done, but you go, you go dig. What she found was that because the high-performance excavator required very little stimulus to get to work digging, they dug about the same on their own as in a group. But the ants who could put off digging until tomorrow, when paired with a diggy-digger ant, actually cut back on their spadework.
This applies to dishwashing and other household chores--people-wise, not with ants--because typically one person in a couple has a lower threshold for the untidy. They notice untidiness before the other one does, and react. But what this means is that they are stuck tidying up because the dishes never fill up the sink and overflow onto the floor, perhaps triggering a response from the other person. There's not even tit-for-tat reciprocity, because if the work gets done before the other person notices it needed to be done, it never registers that it got done.
So there you go. If you clean up all the time, you're demanding to be the cleaner-upper. But note that ant colonies are not made solely of high-performance excavators. Why? Because they die faster than ants that sit on their ant-ass at home.
Fewell also talked about bees, which are more photogenic than ants. That's our feeling, not hers. We're just explaining the photo selection.
Now, a few bee-findings: when pollen-gatherer bees come back to the hive, they do their "Hey, sweet pollen, dood!" dance to non-pollen-gatherers. Similar reason as with the ants. Pollen-gathering bees don't need a dance to get them outdoors. Observing this, Fewell points out that communication in bee colonies is between task groups, not necessarily members of the same group. She says it's a parallel-processor model--strong bridging ties between subgroups allow for fast communication. Otherwise, everyone keeps their bee-nose to the bee-grindstone and focuses on what they do best.
Sean O'Donnell's presentation was built around a model he's constructed to illustrate how the design of communications systems affects group behavior. He didn't actually talk that much about insects until the Q&A, when he went off on how paper wasps bite each other to signal that more foragers are needed. Initially he thought the biting was some dominance display, but the more he watched, the more he became convinced that while it may have been a reproductive-fitness behavior at one point, biting was strictly a communicative technique. Nice, huh? Fucking wasps, man.
On to the modeling! O'Donnell is exploring modularity in a network, so his models have to do with probability-based changes of state, and include time-effects (immediacy and persistence). In practical terms, modular network effects are seen in advertising (where frequency and repetition play a role, but don't guarantee any one person will buy your soap), courtship (which can also involve judgments about frequency and persistence and someone buying your soap), and, in insect life, in recruitment of workers to kinds of jobs.
His key variables are group design, behavior, and cognition, and he's been mapping out how single variable adjustments can result in immensely different kinds of network behavior. His most surprising finding, running the model, was that varying cognition (whether or not an node in a network "remembers" its previous states) led to highly unpredictable behavior. "Forgetting" can be toxic to a network's communication--not in that it prevents responses, but because the network can't respond consistently to a consistent stimulus.
