What's a physicist to do?
One of the things I want to use this blog for (when I actually get around to posting to it regularly) is writing down ideas I have in order to clarify them to myself; I have a feeling they'll start off rather incomprehensible, for which I apologise... hopefully later posts will make things clearer as they develop in my mind. So... with that said, I have an idea I'd like to work through today...
One of the dangers a physicist faces coming over to biology is the tendency to want to generalise: biologists hate it, and in most cases, the validity of the results you obtain is questionable, or far more limited than expected. (And you can see my leanings --- I'm already generalising... sorry to any biologists reading this :) ) Typically, physical science types try to justify their generalisations using teleological arguments: we assume that a particular biological structure has a particular function, and then try to show that it is optimised for performing that function. This is another thing biologists hate: I'm not sure on the history, but the impression I get is that in the past, people had put a lot of emphasis on teleology in biology --- i.e. life was thought to have meaning, and hence `god-given function' --- and letting go of that focus was quite a painful process. Hence, using that kind of argument is a bit dangerous in biology...
I don't think that teleological justifications are incorrect by the way; along with many other folk, I think that the theory of evolution gives it some mileage. As a physicist, I essentially think of evolution as an optimization process. Similarly, as engineers know, a very good way to go about designing something is to come up with a series of components to achieve certain aims in the structure, and then optimize each component to perform its alotted task (within the constraints of money and time... :) ). So at least at some level, by approaching an organism as a `designed' system, we can take parts of it and treat them as components with particular functions. The assumption is then that at least for the `mission-critical' components --- i.e the bits that are really, really necessary for keeping you alive! --- evolution leads to optimised function.
Of course, there are many difficulties with this approach. Unlike human-designed systems, biological systems can rarely be divided up into neat components with particular functions. One great analogy I heard is that evolution is not really an engineer, but more of a hacker --- she takes the `design flaws' in a particular component, and puts them to work in an entirely unexpected way... (There are some amazing examples of this in biology, but I can't remember them off the top of my head... I'll include them in the next version of this post...)
So, where am I going with this? As a physicist recently come into biology, I obviously want to be able to actually do research that is not plain trash, based on questionable assumptions. That leaves me with two ways to go: studying particular systems in detail (for instance, the brains of mosquitos...), with no attempt to generalise, OR finding `handles' that allow me to justify generalising a finding in one system to others. I don't find the first choice particularly appealing --- I studied physics because I was interested in the big picture --- how does the universe work??! So studying a single system in excruciating detail (without planning to extend the results to other systems) doesn't interest me in the slightest (well... with some exceptions). Which leaves me the option of trying to find ways to justify generalisations --- I could go with the `evolution optimizes, so let's try to work out the optimal way of doing this...' approach, but I feel that that is questionable in many instances, as I said above.
So what I'd like to do is to work out what the structures are in biological systems that allow generalisation. I guess the whole thing comes down to a specialised form of epistemology --- in physics, we tend not to worry too much about the fact that our theories are really all wrong; it's just a fact of life, and they do a pretty good job, so we leave worrying about such things to the philosophers. In biology, however, I think the situation is a little more subtle --- evolution gives us some handles on generalising results for biological systems, but exactly how we can generalise is not obvious.
One of the dangers a physicist faces coming over to biology is the tendency to want to generalise: biologists hate it, and in most cases, the validity of the results you obtain is questionable, or far more limited than expected. (And you can see my leanings --- I'm already generalising... sorry to any biologists reading this :) ) Typically, physical science types try to justify their generalisations using teleological arguments: we assume that a particular biological structure has a particular function, and then try to show that it is optimised for performing that function. This is another thing biologists hate: I'm not sure on the history, but the impression I get is that in the past, people had put a lot of emphasis on teleology in biology --- i.e. life was thought to have meaning, and hence `god-given function' --- and letting go of that focus was quite a painful process. Hence, using that kind of argument is a bit dangerous in biology...
I don't think that teleological justifications are incorrect by the way; along with many other folk, I think that the theory of evolution gives it some mileage. As a physicist, I essentially think of evolution as an optimization process. Similarly, as engineers know, a very good way to go about designing something is to come up with a series of components to achieve certain aims in the structure, and then optimize each component to perform its alotted task (within the constraints of money and time... :) ). So at least at some level, by approaching an organism as a `designed' system, we can take parts of it and treat them as components with particular functions. The assumption is then that at least for the `mission-critical' components --- i.e the bits that are really, really necessary for keeping you alive! --- evolution leads to optimised function.
Of course, there are many difficulties with this approach. Unlike human-designed systems, biological systems can rarely be divided up into neat components with particular functions. One great analogy I heard is that evolution is not really an engineer, but more of a hacker --- she takes the `design flaws' in a particular component, and puts them to work in an entirely unexpected way... (There are some amazing examples of this in biology, but I can't remember them off the top of my head... I'll include them in the next version of this post...)
So, where am I going with this? As a physicist recently come into biology, I obviously want to be able to actually do research that is not plain trash, based on questionable assumptions. That leaves me with two ways to go: studying particular systems in detail (for instance, the brains of mosquitos...), with no attempt to generalise, OR finding `handles' that allow me to justify generalising a finding in one system to others. I don't find the first choice particularly appealing --- I studied physics because I was interested in the big picture --- how does the universe work??! So studying a single system in excruciating detail (without planning to extend the results to other systems) doesn't interest me in the slightest (well... with some exceptions). Which leaves me the option of trying to find ways to justify generalisations --- I could go with the `evolution optimizes, so let's try to work out the optimal way of doing this...' approach, but I feel that that is questionable in many instances, as I said above.
So what I'd like to do is to work out what the structures are in biological systems that allow generalisation. I guess the whole thing comes down to a specialised form of epistemology --- in physics, we tend not to worry too much about the fact that our theories are really all wrong; it's just a fact of life, and they do a pretty good job, so we leave worrying about such things to the philosophers. In biology, however, I think the situation is a little more subtle --- evolution gives us some handles on generalising results for biological systems, but exactly how we can generalise is not obvious.