Thursday, November 4, 2021#
Plan#
Going to try Pomodoro today when I get into my first main task, won't use pomodoro for just planning, emails etc. though.
- Read last papers on my next project and finalise some ideas.
- Revise the connectivity paper, and consider how best to link to 1.
- If have extra time, clean up emails, GitHub and notes - start integrating into thesis.
Time worked#
- 3 pomodoros on task 1, getting some decent ideas now. Read papers and found many ideas about correlation.
- 1 pomodoro on task 2, need to get the latest version from Shane.
- Task 3 - 2 pomodoros
So overall, 3 hours in pomodoros, but 4.5 hours by work timer. So haven't really lined these up with my actualy working.
Notes#
Kording 2011 - "In most cases, encoding and decoding models are tightly linked; leading decoding models are usually based on explicit models of encoding"
Kording 2020 machine learning in NS paper interesting point, that the stats algorithms would be better than creating custom algorithms for neuroscience.
Perhaps the same could be said about the analysis I want to perform.
Mathematics and statistics have been around for a long time. They had the answer to my first question in a very simple manner.
As such, perhaps they hold the answer for my later questions as well.
With this in mind, developing new algorithms would not be overly useful, which makes sense. Instead the aim should be to use the existing mathematics in a sensible manner to answer my questions.
This could be particularly important if I choose to look at correlation Correlations enhance the behavioral readout of neural population activity in association cortex | Nature Neuroscience