I was a neuron once
Two baby giants met last evening, 23rd of November, to play Pong against each other at the Champalimaud Foundation (Figure 1). By waving colored cards at a camera’s eye, each person in the auditorium had, for a few minutes, the privilege of being a neuron in the mind of one of these giants, making decisions about where to move the paddle. An auditorium split in half, turned into a pair of 200-headed monsters. A final score of 13 – 6, a flawless victory of the left wing (no political connotation implied, obviously).
Multitudes of individual elements self-organize and an interesting phenomenon emerges to a bird’s-eye view observer. Local rules giving rise to complex large-scale patterns, the big as a consequence of the small – no overarching organizer agent needed. A theme with many variations in nature, as clarified for us by the Champalimaud young neuroscientist Tiago Marques. From atoms interacting to form molecules to humans interacting to form societies and, somewhere in between these levels, cognition emerging from local neuronal interactions, a subject very much dear to our neuroscientists’ heart.
Figure 1. Two 200-headed giants were seen playing Pong at Champalimaud Foundation this week. I confess I was a neuron making up the mind of one of them.
Butterflies in the playground
Manuel Marques Pita, a computational biologist at Instituto Gulbenkian de Ciência (IGC), brought more kids into the playground. They had their own colored cards but a different game: they would pick the color of their cards depending on the color of their neighbors’, as dictated by some very simple rule, until everyone had the same color. From this childish-looking game, we drew the following insights: 1) for network-wide patterns to emerge, information has to be able to propagate over large distances; 2) games that look very distinct at the onset can look the same after some time, so it is impossible to reverse-engineer the initial conditions; 3) interesting patterns also emerge at intermediate levels (i.e. smaller than the whole network but larger than the elements); 4) small changes at the onset can have a huge impact in the outcome of a game, a phenomenon known as the ‘butterfly effect’.
Speaking of butterflies, Manuel finished his talk showing us how the colorful patterns on the butterfly wings emerge from a game played by cells listening to their neighbors and deciding which genes to express. Those were impressive images. Someone in the audience pointed out yet another instance of the cards games: perhaps if we just replace “pick the color of your card depending on the color of your neighbors” by “define your political decisions and positions depending on those of your neighbors”, the same rules would apply to politicians at an Assembly.
Figure 2. The legend goes that a flap of a butterfly’s wings can trigger a hurricane – a metaphor of complex systems’ sensitivity to initial conditions.
Disparate entities, common patterns
At the next talk, no babies or kids, but teenagers. José Leal, head of the Computational Genomics Lab at IGC, explored the similarities between social networks and protein interaction networks. He brought us to a deeper understanding of network functions by showing us that besides being able to learn about large scale dynamics if we know the local rules, the opposite is also possible: we can learn about the local rules themselves by studying the large scale patterns. As an example (a quite entertaining one), he showed a social network depicting who had had sex with whom in a group of high-school students. Interestingly, a social local-interaction rule popped out: teenagers tend not to swap partners (i.e., “love squares” are rare), a rule that doesn’t hold so well for adults. Sexual culture in adolescents, protein interactions – “How is it possible to learn anything about commonalities from such diverse things?”, one might reasonably ask. Zé Leal’s message, however, is clear: “we study interactions, not entities”. And interactions, unlike entities, don’t care about size.
The Queen’s anarchy
To close the evening, Deborah Gordon introduced us to a kind of social network we don’t tend to engage in (Figure 3). Colonies of harvester ants consist of a fertile queen, a few small-headed males (who do nothing but copulate and die – they barely get to eat in their short lifetimes!) and many sterile female workers who do all the jobs (cleaning, caring for the baby-larvae and searching for food). While searching for food, ants are exposed to predators and sunlight, facing the risk of drying out. However, if they don’t forage enough, their territory might be taken over by a neighboring colony and the colony will ultimately starve. A balance needs to be reached, although there’s no manager ant in charge of it – the balance emerges from local interactions between the ants.
Figure 3. As Deborah (left) explained, forager ants decide whether or not to go out of the nest based on a rough count of their returning fellows. This sort of local information ultimately determines the foraging policy of the whole colony.
Deborah’s talk stressed an important aspect of emergent properties: they result from interactions between agents that only have access to local information. Like Manuel’s children playing the color cards games, or like politicians or high-school students establishing a culture, ants need not be aware of what goes on at the global level. In fact, Deborah told us that, when deciding whether or not to leave the nest, all an ant needs to know is how frequently the other ants are returning safely and bringing food. How does Deborah know that? That’s easy, she has a plastic box! Every time an ant returns to the nest with food she traps it in her plastic box, and as a consequence ants in the nest will refrain from leaving. However, if little glass balls with ants’ smell are dropped in the nest entrance, the ants will leave the nest as if other ants were returning. From these experiments we learn that forager ants decide whether or not to go out of the nest based on a rough count of their returning fellows. This sort of local information ultimately determines the foraging policy of the whole colony – curiously, pretty much the same way as data flow over the internet is controlled by the TCP protocol.
A theme emerges
Local rules giving rise to complex large-scale patterns. The big as a consequence of the small. Ant colonies, internet, social networks, butterfly wing development, a brain making decisions. Variations of Emergence, a theme we are glad to have explored, and we hope you enjoyed as much as we did.
Thiago Gouvêa believes in social self-organization