My Ph.D. in plain English…

I guess that this meme has been going around on Twitter for a bit – I picked it up over at Carly Tetley’s blog Wildlife Research and Training – but I thought that since I had submitted my thesis to the department for review (avant de la défense), it was a good time to write about my Ph.D. work in plain English.

My Ph.D. research has been about the evolutionary foundations of social foraging behaviour in animals.  What does that mean?  Well, social foraging is the study of foraging decisions that animals make when they’re in groups, and when the decisions that they make depend on what the other members of the group are going to do.  This is an inherently game theoretical problem.  Now, that won’t mean much to you unless you know what game theory is, so here’s an illustrative example:  imagine that you’re at a party, and you get snackish.  You look over and see the snack table loaded with all sorts of goodies, from cookies to cakes and everything in between.  Individual (a.k.a. optimal) foraging research would study your decision of what snack to take based only on what snack you prefer.  Social foraging research would consider your decision-making process when you and your two best friends head to the snack table at the same time.  If all three of you like cookies, and there’s only two cookies left on the table, then it might be a smart decision for you to switch to cake – even though you prefer cookies over cake – rather than engaging in a bare-knuckle brawl over the last piece of chocolate chip heaven.  We can apply the same logic to the study of animals foraging and interacting in groups.  (If you’re paying attention, you might notice that there’s a third possibility where individuals forage in groups but make decisions independently;  this scenario corresponds to the outcome where everyone at the party has their own plate of goodies to choose from.  You forage together, but your decisions don’t affect each other).

Birds foraging socially...

A slide from my Ph.D. seminar: birds foraging socially.

We know that a lot of species across many taxa forage socially; for instance, it has been observed in birds, fish, mammals, and there’s even evidence for insects and possibly bacteria.  In these foraging species, the most common social foraging game observed is what’s known as the “Producer-Scrounger game”.  This is a game in which individuals take one of two roles, as the name suggests:  producers or scroungers.  Producers spend their time searching for food resources, while scroungers wait for a producer to find a food resource and then they join in the discovery.  Extending the party metaphor above, if you were producing you would be searching through the room to find a table with food on it;  a scrounger would be that lazy friend who waits for you to do the work of finding the goodies before strolling over to take advantage of your effort and help themselves to whatever’s on the table.   In foraging systems, there will be an mix of these two tactics where the “fitness” (usually measured by proxy as food intake, i.e. the number of cookies you scarf) of the two are equal.  This is what’s known as an ESS, or evolutionarily stable strategy.  I don’t want to delve too deeply into evolutionary game theory here, but you can think of the ESS as the best mix of producing and scrounging for you to play given the mix that everyone else is playing.

That’s the back-story to my Ph.D.  My research has focused on the theory of these social foraging games, and how to extend them to match real foraging situations more effectively.  For instance, most of the work done on the producer-scrounger game to date has been very agnostic when it comes to representing the world spatially.  This is deeply weird to me, because if you spend more than a few seconds looking at animals foraging in the wild it becomes obvious that spatial relationships – both between foragers and between foragers and their environment – have a significant impact.  Close foragers will interact more heavily; a patchy, broken landscape will be different to forage on than a regular grid with patches spaced evenly;  and so on.  Adding these spatial components into the theory of social foraging has been a major focus for me.

The other major theme of my thesis has been information use.  In behavioural ecology, “information” has a specific meaning that relates to how animals use observations of the world around them, especially other animals, to make decisions.  In foraging terms, this often works out to “Hey, how is Bob getting along at that patch over there?  Oh, he hit the jackpot!  Let’s go get some of that!”  Anthropomorphism aside, we can ask sensible questions about how animals collect and use public and private information.  Glossing over some nuances, we can think of private information as information gathered by the animals itself and not accessible to any other observer, like information about the richness of a patch gathered by sticking your head into it.  You can see what’s in there, but no-one else can.  Public information, on the other hand, is information that is accessible to anyone who’s paying attention.  If I’m a producer who has found a food table at a party, this becomes obvious to anyone tracking my movements when I begin stuffing cookies into my mouth as fast as I can.  Scroungers rely on public information to scrounge, otherwise the game would break down;  this means that information use is central to the study of social foraging.

For historical reasons, though, behavioural ecologists haven’t spent much time thinking about the mechanisms by which animals use this information.  They’ve vaguely assumed that natural selection will have worked this out, but haven’t done much to figure out what that product will be.  In social foraging, it has always been assumed that natural selection will bring animals to the producer-scrounger ESS (the optimal foraging strategy) on its own.  But we see animals adjusting their use of the producer and scrounger tactics over their lifetime, and often on a very short time scale (seconds, not generations) as they respond to rapidly changing environments. So how do they do this?  I’ve spent a fair bit of time looking at mechanisms that will allow an animal to learn an ESS, and how natural selection might act on those mechanisms instead of fixing an ESS right off the bat.

Answering these questions, both about space and learning, has required the use of computer simulations to augment the mathematical models that currently exist;  unfortunately, creating new formal models of these processes is an extremely difficult task and I prefer to let the computer do that work for me.  Therefore, I’ve spent a lot of time creating individual-based models and genetic algorithms to study these questions;  in the interest of keeping this post to a reasonable length, I’ll refer the interested reader to the Wikipedia pages for those topics, and I would be happy to answer any questions in the comments.

And I could talk about this for hours, but I think I’ll cut off the level of detail there so that I don’t drown innocent readers in progressive elaborations.  In any case, that’s a high-level view of the type of research that I have been involved in for the past four years.  Please feel free to ask questions in the comments!

Tagged ,

2 thoughts on “My Ph.D. in plain English…

  1. I find game theory fascinating and I just got all excited about agent based modelling when I went to the ASAB Easter meeting. They did try to teach me how to write models in R in my first year as a PhD student, but it was slightly too abstract for me. I think that’s probably why I got so excited about StarLogo. Created for children. Sweet 😛

    Do you write all the code for your models yourself?

    • stevenhamblin says:

      You know, I *love* R and use it every day. I also *hate* R, and I would rather smash my own brains out with a shovel than use it as a first language when teaching people programming or modelling. Short of writing it in Brainf@#k, I can’t imagine too many worse ways to introduce people to computational modelling.


      To your question, though, yeah I do write all of my own code. I write much of it in Python, which is a language I would recommend for newbies. It’s easy to read and understand, while maintaining as much power as you’re likely to need. I’ve also written models and code in a half-dozen languages recently, though. I like writing my own code, because I always feel that I understand the models much better if I’ve gotten into their guts rather than letting a library do it…

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s