My Feed

Tue, 10 Sep 2024 13:44:48 GMT

I am more than pleased to share my first, first author pre-print with the irrepressible @maxdiluca & @Jennifer_L_Cook 'Biological kinematics: a detailed review of the velocity-curvature power law calculation' https://doi.org/10.31234/osf.io/vfq3d. A short thread on the paper & a story of how an Electronic Engineer ended up doing a #PhD in Psychology!

The velocity-curvature power law is ubiquitous in biological movement & deceptively simple; tight curves, slower movement! Lacquaniti et al. in 1983 had found an empirical power of ~1/3 linking curvature & velocity. It's since been hailed as a kinematic law of nature, potentially underlying all biological movement? Any candidate fundamental law should have a vetted, open source, method of calculation. The lack of such a toolkit, we aim to remedy with our paper.

Often research is a Garden of Forking Paths, giving variable outcomes for differing analyses, but in the power law's case the 1/3 value has enormous gravity. Each step in the power law calculation– filtering, differentiation & regression – can potentially bias towards 1/3. We distil a toolkit, building on key insights from the literature, & demonstrate it does not fall foul of this bias; https://github.com/dagmarfraser/velocity-curvature-power-law-protocol

The code & discussions are relevant to all #kinematics research using zero lag 10Hz low pass Butterworth filters, using finite differences differentiation (velocity to acceleration to jerk), or employing linear regression after a log projection in the presence of measurement uncertainties.

If you made it this far.. you are probably wondering how did an Electronic Engineer end up as a #Psychologist? I blame #MATLAB. I first encountered MATLAB in an asylum! @EdinburghNapier's Craiglockart campus is the former WW1 military psychiatric hospital, where the shell-shocked Sassoon & Owen penned some of their most famous works.

It's summer '98 & I'm a newly minted BEng(Hons) Electronic & Communication Engineer quantifying filters for CCD cameras… in #MATLAB! It's completely new to me, but it's like programming BASIC on my ZX Spectrum+ in my bedroom in the 80's. A sense of play. The immediate feedback from the IDE doled out the perfect dopamine hit.

Now it's 2007 & I am just back from being a seasonaire in @Courchevel & I need a 'real' job sharpish, but I recall that halcyon summer of '98, so I type #MATLAB into jobs.ac.uk – & hit enter. The top hit is Sensory Motor Neuroscience Lab at UoB (hey @dr_mte). Wait, do they even use #MATLAB in Psychology?

Transpires the PsychoPhysics Toolbox (@psychtoolbox, send them money! https://xkcd.com/2347/ ) transforms a humble desktop PC into a near perfect psychology experiment hosting battlestation. Thus, you can run the experiment in #MATLAB & analyse the data there too. The perfect playground for me. They paid me to do this!

It's 2018 now a decade+ of experience coding MATLAB experiments & analysis suites for Psychologists. I have made every possible mistake at least twice. @Jennifer_L_Cook hires me as an iOS Programmer, but we get sidetracked when I foolishly say I can reimplement Fig. 8 from Huh & Sejnowki's 2015 Spectrum of Power Laws for novel data, from autistic participants.

It awoke half remembered knowledge of FFTs & filters from my Elec Eng undergrad and gelled with everything from the last decade. Deeper & deeper down the rabbit hole I went. We recognised the need for the vetted toolkit to tease apart the effect of analysis choices from the underlying data, so in 2022 an Electronic Engineer starts a PT PhD in Psychology because of #MATLAB. This is probably not the most directly useful career advice but MATLAB can open doors!

Tue, 10 Sep 2024 13:44:46 GMT