3/15/17 Oops! I’ve been completely forgetting to start a new doc for 2017 and update this journal. But then again things have been so complex and mentally challenging for the past few months that I wouldn’t have been able to figure out what to say. You should see my reams of notes! Today I feel like I’m moving forward a bit, anyway. Progress on all the fronts that have been bugging me so much: executive function, short-term memory, emotional processing, generalization… And also the feeling that I’m on to something in the whole consciousness arena, regarding what I’m calling the ‘virtual now’. It’ll probably all go wrong again tomorrow but the final structure is starting to reveal itself, I think. Brains are remarkably holistic systems and it’s been like trying to push down bubbles in wallpaper - get one flattened out and three more pop up somewhere else. I’ve never thought so hard in my life. Funny to think I just set out to write a video game…
3/29/17 I’ve gone more than twelve hours without my latest idea falling over. Wondering what will ruin it this time.
4/11/17 So far the executive ideas are panning out, I think, apart from a few loose ends. I’ve coded it but can’t actually test/debug anything until I build some more brain maps and instincts, to generate a comprehensive action repertoire. I’m trying to resist having to build the visual brain mapper yet to find my way around, access annotations, etc, because that’ll be a lab instrument for users and so needs a lot of thought, although things are getting pretty complex now - there are 120 maps and a lot of connections between them. Instinct patterns are the worst - it’s hard to visualize and edit them. But I mustn’t get sucked into making tools right now.
4/14/17 Jeez! Real maps of the brain are starting to look simple by comparison! Today I’ve mostly focused on how executive cortex can make plans that involve changes of state in objects that are out of sight and hence only known about from short-term memory. Including the ability to imagine walking up to something (hence altering its hypothetical proximity status and providing feedback about how tired it’ll make us, or whether we have to go through the scary forest to get there). Short-term memory substantially changes the way my theory has to work above that level, but there’s no mental life without it, so I just have to work through it bit by bit. Once I’ve finished that, the whole executive system is basically finished. None of it will work, of course, and debugging it is going to be absolutely horrendous, but it’ll be a huge weight off my mind to have it written. Apart from any language stuff and finalizing the affective system, that will mean the whole brain is finally coded into the genome (currently nearly 600 gene pairs!). One thing I need soon, though, is enough for the creatures to DO, or it’ll be hard to know if they’re learning properly.
4/24/17 Somehow I got sucked into doing language. It seemed kind of handy to have gloops be able to tell me what they’re thinking or doing, but there was a lot more to it than I expected. Still, it needed doing eventually, I guess…
5/5/17 Ploughing on. Associative learning, basic language, and a gradual firming up of executive function. Still bothered by the relationship between conscious planning and unconscious sequencing. I don’t have it right, somewhere. But if I can find a way to integrate object status and object-directed actions that’s more ideomotor, I might be onto a way forward. If so then that’s the brain basically built, apart from beefing up the limbic stuff. Debugging it will be, er, interesting.
5/19/17 Almost ready to debug the code I wrote a month ago!
5/23/17 Furiously debugging. They’re not easy bugs to find, though, in such a massively parallel system. I alternate between using a debugger, and using the equivalent of an fMRI scanner and neurelectrodes. Gradually pinning it all down. Mostly it’s a question of why various cortical maps don’t obey the yin-yang rule, or deal properly with imaginary signals, etc.
5/26/17 If I was being paid to do research, things would be going great! I learn so much from getting things wrong. All sorts of insights into attention, learning, etc. that probably nobody else knows. But right now I need to be getting things RIGHT.
6/2/17 Getting close to being able to resume up-to-date builds, at long last, but a) I need to change the way signaling and visual appearance work, b) I need to add some more emotional processing (hungry, cold, wet, etc. are about all we have, even though the base chemistry is there for more) and c) gloop currently just lies there doing nothing, so it’s really not much use uploading it until he’s motivated to get off his ass and poke around a bit. Even then there’s very little for him to do and there are tons of bugs in the executive systems, but at least it will actually run in standalone form once more. Shouldn’t take more than a few days.
6/9/17 MAJOR overhaul of perceptual properties. Mostly done but it’s likely to have broken things. I think it was worth it now that I can see the common themes emerging in perception and perceptibility.
6/16/17 I find myself fiddling with visual saccades yet again! Eye movements and attention in general are turning out to be very important, though. Just got to incorporate affective perceptual properties and produce some limbic maps and then maybe gloop will get up off his ass and start doing things again.
6/23/17 I really wish I’d tried to get a research fellowship or something to do this. I’m still having SO many new ideas and insights, and in a research context this would be fantastic - each one a new paper. But as far as product development goes, I just need answers, pronto, not whole new avenues to follow up. Then again, half the reason I’m sitting here on my own in America is because it was a fucking nightmare trying to get people in a traditional context to support my work. Way too ambitious and flaky for academia, way too risky and (ironically) unambitious for business grants. I wasn’t interested in IPOs and ROI, so the business people wouldn’t touch me - nobody’s interested any more in businesses that actually make things people want to buy and then make a profit from the sales. And academia claims to want ambitious, high-risk ideas, but they don’t really. They want people who can find grants to pay PhD students. If I gave this problem to PhD students, what would be the point of my 40 years of unique experience and intuition? I can’t teach this stuff to people - it’s too intuitive - so I’d just be the one writing the grant proposals while other people got no further with it than anyone’s doing already. There’s no shortage of people already working on this stuff. So it never happened and was part of the reason I had a crisis. So instead I’m broke, worn out, and desperate for the new ideas to stop, so that I can actually finish it. I don’t want to be standing at an intriguing fork in the road - just a nice one-way street with no choices to make!
Having said that, this week’s gone pretty well. Saccades are much better and I have a comprehensive (and genetically tweakable) system for handling perceptual features. Also a working draft of the assorted ‘limbic nuclei’ needed for turning perceptual features into moods, drives and emotions. But I need to finally pin down WHICH features and exactly what the emotional rules are. It was going great until I started to think about friends and enemies, how we make them and what perceptual structure we need. And that’s created a new damn fork in the road!
7/7/17 At last Gloop is up and about, instead of just lying there and only understandable via brain scanners! What he does is often nonsensical and his executive cortex is stuffed full of bugs, but now that he’s active again I can go ahead and start cutting regular builds. I don’t think things will break as badly any more. I just want to add some more emotions and drives and then I’ll see if it works when I do a standalone build. It’ll be ready some time next week, I’m pretty sure.
BTW, huge thanks to those of you who donated some extra cash! You know who you are.
7/14/17 I was SURE I was going to get a build out today but at the last minute I’ve broken the brain somewhere. Weird stuff is going on and it’s going to take some patience to sort out. I must get it built by next Friday, though, because I’m going away for a few days. Damn. Sorry.
7/20/17 Damn, damn, damn. Not going to happen. The executive map needs to be rewritten. I learned a lot, but after patching the patches of the patches it became obvious I’d taken the wrong approach. I have a better idea but I’ll be away for a week starting tomorrow, so it’s going to be delayed until a few days after that. It was nice to see gloop look at me and actually think what to do about me for a while, but now it’s broken again until the rewrite. So close, though.
9/7/17 I’m STILL stuck on the whole planning and executive problem. I’ve tried SO many different solutions. I know it can be done, but every solution I’ve come up with comes unstuck at some point that I couldn’t see before I got to it. It’s just the way with inventing things, but I really need an answer soon. Yesterday, out of the blue, I came up with yet another approach and it sounds promising (although I’ve said that before!). So many of the hard parts were because I was trying to do the planning part the easy way, so I’ve turned this on its head and now the hard parts are easy but the planning part is hard. I MIGHT have found a way to do it, though. It’s quite a nice, dynamic way. But we’ll see. It’s so radical compared to all the other methods that my mind isn’t conditioned to think this way yet.