A more serious attempt to re-vision a personalized "congitive basis"
It all started as an involuntary study induced by immense cognitive load of a graduate school back in 2016.
The study core essence stayed the same - searching for the subjective choice of cognitive "modes" that form a complete basis.
Now it seems I ended up in the domain of cognitive topology, benchmarking AI models reasoning purely for fun!
I wish that would provide me some financial support as well ^_^
But hey - at least I can look down from my moral highground on this AI-obssesed 10x-productive crowd
until I probably die of starvation in my isolated bubble inside ostracized country
!
Caution
Obviously, I do not claim presented here as a legit scientific research. The latter is what I do on my day job for more than a decade now (with a very embarrassing h-index), so naturally I'm against pseudoscience. Instead think of this repository content as a philosophical exercise - I am trying to understand my own (subjective) cognition. But at the same time, it is interesting to analyze how objectively reasonable it may be generalized?
| Explora∫n | Defini∫n | Negotia∫n | Specifica∫n | Genera∫n | Revela∫n | Integra∫n |
|---|---|---|---|---|---|---|
| Where? | When? | Who? | Which? | How? | Why? | What? |
| Sunday | Monday | Tuesday | Wednesday | Thursday | Friday | Saturday |
The subjective benchmark - a sparring match with an LLM
This time I'm rebuilding a much tighter foundation from the ground up by playing an intellectual (word -> description) game with various Large Language Models (LLMs). No extra metaphorical anchoring and symbolism and less cultural noise overall to try distill the reasoning basis first. Rather emotional tone setting - I admit it, but I'm not a robot. I would even argure that adding this emotional side taps into the model emphaty but the guideline in this prompt does explicitly asks for the disabling personal glazing and directing attention towards ideas first. I am not very serios about developing this to the point worth submitting in a peer-reviewed journal (yet) - mostly because I don't want to degenerate to the level of pseudo-science and actually researching this to the level required would definetely require finding a co-author who is a specialist in this field. But it is just a good practice to do meticulous work documenting anyway - who knows how this non-profit side-project research will turn out? Maybe another (better) talk on some DataFest, maybe it would just stau some weird repository which is at least has something original.
You can still check out an even more subjective legacy version (with a tonn of symbolism) in the separate repo branch that was co-designed with DeepSeek-V3 and presented at the DataFest in May 2025.