[Comp-neuro] Re: Attractors, variability and noise

Axel HUTT axel.hutt at loria.fr
Fri Aug 22 14:15:38 CEST 2008

> One thing that has always struck me about the difference between  
> biology and physics, is that in physics department seminars, the  
> resident Dons of the field  would stand up and walk out within several  
> milliseconds of a speaker making a statement violating some well  
> established physical principle (conservation law or something), while  
> in biology, one finds oneself listening intently through most seminars  
> to determine whether the story teller does or doesn't know what they  
> are doing.  There are little signs (smiling gels, action potentials  
> that don't overshoot, etc), but really one is often at sea.
> So, there is clearly something different about the freedom to tell  
> stories in both disciplines.
> In biology, when you work on a new paper, the first thing you do is  
> organize the figures (the story). In math you organize the equations.   
> In biology the traces you pick for your figures are the best ones you  
> can find (and photoshop can help) whereas physics and mathematics look  
> for general solutions.
as a physicist, may be I am allowed to say that most people working
seriously in classical topics of physics (standard topics taught at
university like quantum theory, relativity theory, solid state physics,
at the end everything not related to chemistry or biology) are
religious fundamentalists in the religion physics as a religion (and we
are back at the tyranny of ideas mentioned previously in the
discussion). However physicists with interest also in other fields, 
such like biology, chemistry, computer science or human sciences
are more relaxed and give other people more freedom in their ideas.
Since this is the case for most neuroscientists, I would expect (or
hope) that there are only few physicists like the ones you described. 

Similar to your example, recently I (physicist by education) had a very
interesting discussion with a colleague (cognitive neuroscientist  by
education) about subtle differences in our underlying research
approaches in neuroscience. Either of us aims to model mathematically 
neural dynamics but:
she takes a close look of the system under study (e.g. the basal
ganglia), aims to grasp its major properties (e.g. spiking dynamics
under different experimental conditions, axonal connectivity
structures, neuron types etc), takes a model, puts in all knowledge
(e.g. parameter ranges) as far as possible and studies the model 
under these constraints to match the model results to the knowledge 
obtained previously. IMHO a great approach.
In contrast, my research considers rather general models, which imply
all possible parameters and connections. The outcome of my research are
conditions on the parameters, connectivities, neuron types etc. under
which one finds a specific behavior. The specific behavior is taken 
from the literature.Hence I could argue that my approach is more general
and might be applied to many different brain areas in different species
in different experimental setups. However, I guess her results are 
a bit closer to the real neural behavior.

Although the approach of my colleague and mine essentially may aim to
describe the same experimental phenomenon, we choose rather distinct 
approaches. At the end of the discussion, we agreed that our own
specific interests decides on the approach.

> This is my problem with all for one and one for all -- somewhere,  
> somehow, someway, neuroscience has to find a mechanism to better  
> discriminate between efforts -- let alone force the development of  
> some set of common definitions with which and on which to tell stories  
> (and judge stories).  Otherwise, they are just stories and subject to  
> all the politics, trends, fads, and social pressures that apply to all  
> story tellers.  So a core part of this debate, I think, is how we  
> advance the 'art' of storytelling in biology.

in my experience a good way to solve the problems is the choice of a 
specific mutual question, which is studied by different groups.
For instance, several years ago I was very interested in the origin
and detection of cognitive components in event-related potentials (EEG
measured during cognitive experimental paradigms). I have observed
many different approaches in the literature for their classification:
the neuropsychologists classify single time series them by their
latency, height, duration and sometimes their spatial location on the
scalp. Computer scientists have detected them by multivariate 
data analysis tools (e.g. developed by D.Lehmann and P.D.
Pascual-Marquis). I have taaken their results into account and argued
that the ERP-component may be the result of highly synchronized 
underlying neural activity and thus modeled the multivariate
ERP-components by a set of ordinary differential equations. It turned
out that the ERP-component P30 in middle-latent auditory evoked 
potentials exhibits a unique dynamical behavior (equivalent topology)
in three different subjects. Further one conclusion was that the 
brain exhibits a kind of chaotic itinerancy suspected by many authors,
e.g. as Tsuda or recently Gros. 
I am sure further work on the origin of ERPs is difficult and highly
interdisciplinary but probably fruitful to gain specific insights 
to cognitive processes. All groups irrespective of their education 
may contribute to the understanding as far as one focusses to the 
same question.

Have a nice weekend


More information about the Comp-neuro mailing list