Skip to content

Commit e58c649

Browse files
committed
Hypnagogic Hallucinations blog entry edits
1 parent 513557d commit e58c649

1 file changed

Lines changed: 68 additions & 33 deletions

File tree

docs/pages/aiDev/researchPosts/2026-05-04_A.js

Lines changed: 68 additions & 33 deletions
Original file line numberDiff line numberDiff line change
@@ -6,50 +6,75 @@ entryData.date = "2026-05-04";
66
entryData.eid = "A";
77
entryData.tags = ["theory", "hypnagogic", "hallucinations", "metacognition"];
88
entryData.body = `
9-
While falling asleep last night, had a quite eye opening pre-sleep visual, a hypnagogic hallucination.
9+
While falling asleep last night,
10+
<br>&nbsp;&nbsp; I had quite the eye opening pre-sleep visual,
11+
<br>&nbsp;&nbsp;&nbsp;&nbsp; A hypnagogic hallucination.
1012
1113
<br><br> It was of a red-winged blackbird flying toward me and then perched upon an unseen perch, before disappearing.
1214
13-
<br><br> I opened my eyes and said to myself, "woah, an animated pre-dream"
14-
<br>&nbsp;&nbsp; Before closing my eyes again to see the same bird fly away, with low resolution motion from the wings, "blocky flapping wings?"
15+
<br><br> I opened my eyes and thought to myself,
16+
<br>&nbsp;&nbsp; "Woah, an animated pre-dream"
17+
18+
<br><br> Before closing my eyes again to see the same bird fly away,
19+
<br>&nbsp;&nbsp; With low resolution motion from the wings,
20+
<br>&nbsp;&nbsp;&nbsp;&nbsp; "Blocky flapping wings?"
1521
16-
<br><br> It seemed like the first time I ever saw a hypnagogic hallucination actually move and retain persistence over time.
17-
<br>&nbsp;&nbsp; For me, normally it's a scattering of geometric shapes which form into faces, very many faces, in rapid succession. This was the first time I saw anything cogent enough to visualize as a moving subject in my minds eye.
22+
<br><br> It seemed like the first time I ever saw a hypnagogic hallucination actually animate, move, and persist over time.
1823
19-
<br><br> As I mentioned above, the wings looked blocky. I knew I was seeing flapping wings, but it was as though my convolutional deep-learning graph node of neurons showed itself, specifically the convolutional part, like CNN edge detection within an AI network.
24+
<br><br> Normally for me,
25+
<br>&nbsp;&nbsp; It's a scattering of geometric shapes which form into faces, very many faces, in rapid succession.
26+
<br>&nbsp;&nbsp; This was the first time I saw anything cogent enough to visualize as a moving subject in my minds eye.
2027
21-
<br><br> The blockiness was the tensor of flapping wings of a red-winged blackbird, but the motion itself was the min-max of wing extensions over time. The convolutional edges of time based movement, visualized as a solid shape of "wing", while I knew it was flapping.
28+
<br><br> As I mentioned above, the wings looked blocky; like low resolution animation smearing.
29+
<br>&nbsp;&nbsp; I knew I was seeing flapping wings, but it was as though my convolutional deep-learning graph network of neurons showed itself,
30+
<br>&nbsp;&nbsp; Specifically the convolutional part, like CNN edge detection within an AI network.
2231
32+
<br><br> The blockiness was the tensor of flapping wings of a red-winged blackbird,
33+
<br>&nbsp;&nbsp; Yet the motion smearing itself was the min-max of wing extensions over time.
34+
<br>&nbsp;&nbsp; The convolutional edges of time based movement, visualized as a solid shape of "wing", while I knew it was flapping.
35+
36+
<br><br><br><div class='procPagesAIDevBar'></div>
2337
2438
<br><br> I've been focusing a lot on Time.
25-
<br>&nbsp;&nbsp; Time doesn't look normal in the cases of predicting future states, future word chunks, future predictions.
26-
<br>&nbsp;&nbsp; It operates like a continuum, where sections of time are perceived like seeing a graph of a company on the Stock Exchange.
27-
<br>&nbsp;&nbsp; But in the case of images and multitudinous arrays of data, it becomes a ranges of Edges and shapes.
39+
<br>&nbsp;&nbsp; Time doesn't look like Time when predicting future states, future word chunks, future predictions.
40+
<br>&nbsp;&nbsp; It operates like a continuum, where sections of time are perceived like seeing a chart of a company on the Stock Exchange.
41+
<br>&nbsp;&nbsp; But in the case of images and multitudinous arrays of data, it becomes ranges of edge detections into shapes.
2842
29-
<br><br> I never thought I would have seen my brain visualize its internal edges and shapes in a way I could consciously see what my brain believes motion is.
30-
<br>&nbsp;&nbsp; But then it happened to me last night.
43+
<br><br> I never thought I would have seen my brain visualize its internal edges and shapes in a way I could consciously tell what my brain believes motion is.
44+
<br>&nbsp;&nbsp; But it seemed that happened to me last night.
3145
3246
<br><br> It validated my approach I used in my ESN I have written up on my 'ESN Motion Prediction' page.
3347
<br>&nbsp;&nbsp; The difference is, my context window of motion in that ESN,
34-
<br>&nbsp;&nbsp; Or even the learning-rate over time ESN in my ESRGAN Image Upresser project,
35-
<br>&nbsp;&nbsp; Is the source of a major "complexity" issue.
48+
<br>&nbsp;&nbsp;&nbsp;&nbsp; Or even the ESN I'm using for learning-rate & direction of training epochs in my 'ESRGAN Upresser' project,
49+
<br>&nbsp;&nbsp; This is a source of major "complexity" issues in AI in general.
3650
37-
<br><br> My Motion Detector uses a shifting context window in order to learn what edges exist and should retain after the sliding context window, cyclically, slides over the same area of its brain,
51+
<br><br><br><div class='procPagesAIDevBar'></div>
52+
53+
<br><br> My Motion Predictor uses a shifting context window in order to learn what edges exist and should retain after the sliding context window, cyclically, slides over the same area of its brain,
3854
<br>&nbsp;&nbsp; Over and over again.
3955
40-
<br><br> If that's too cryptic,
41-
<br>&nbsp;&nbsp; There is a bunch of "frames" of data my ESN would fill in,
42-
<br>&nbsp;&nbsp; As motion moves from frame to frame,
43-
<br>&nbsp;&nbsp; The "context window" also moves frame by frame.
44-
<br>&nbsp;&nbsp; But if I have a window of 15 frames, after that 15th frame, it loops back to 0,1,2,3...
45-
<br>&nbsp;&nbsp; So it constantly re-writes prior frames of recorded motion,
46-
<br>&nbsp;&nbsp; Unless the motion found within the last frame to update the current frame...
47-
<br>&nbsp;&nbsp; Like, every 15 frames, it will rewrite the data from 15 frames prior,
48-
<br>&nbsp;&nbsp; Supports the found motion 15 frames later.
49-
<br>&nbsp;&nbsp; Then, motion is retained, edges start to form, and shapes beging to dictate movement on screen.
50-
51-
<br><br> And just to say again, from my Terminology page, my ESN is not a fixed-reservoir, like ESNs usually are.
56+
<br><br> If that was too cryptic,
57+
<br>&nbsp;&nbsp; There is a bunch of "frames" of data my ESN fills in as it watches video,
58+
<br>&nbsp;&nbsp; As shapes move from frame to frame,
59+
<br>&nbsp;&nbsp;&nbsp;&nbsp; The "context window" also moves frame by frame.
60+
61+
<br><br> Lets say I have a window of 15 frames, after that 15th frame, it loops back to 0,1,2,3...
62+
<br>&nbsp;&nbsp; All 15 frames are weighted against each other to find the current frames motion-edges, or the edges of movement.
63+
64+
<br><br> So it constantly re-writes prior frames of recorded motion,
65+
<br>&nbsp;&nbsp; Unless continuous motion is found within the last 15 frames to update the current frame...
66+
67+
<br><br> Like, every 15 frames, it will rewrite the data from 15 frames prior,
68+
<br>&nbsp;&nbsp; If current motion supports the found motion-edges 15 frames later,
69+
<br>&nbsp;&nbsp; Then, motion-edges are retained,
70+
<br>&nbsp;&nbsp;&nbsp;&nbsp; Edges start to strengthen,
71+
<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; And edge-shapes begin to dictate movement on screen.
72+
73+
<br><br> And just to say,
74+
<br>&nbsp;&nbsp; From my AI Terminology page, my ESN is not a fixed-reservoir, like ESNs usually are.
5275
<br>&nbsp;&nbsp; It's a Dynamic reservoir, that reinforces prior found patterns prior to training on the reservoir's output.
76+
<br>&nbsp;&nbsp; The motion-edges I mentioned above are the results of the updating reservoir.
77+
<br>&nbsp;&nbsp;&nbsp;&nbsp; So that, the motion-edges are the "fixed" parts of the reservoir.
5378
5479
<br><br><br><div class='procPagesAIDevBar'></div>
5580
@@ -60,14 +85,16 @@ entryData.body = `
6085
6186
<br><br> If consciousness only starts to take hold while the brain is producing Gama waves,
6287
<br>&nbsp;&nbsp; Yet the motion I saw in the bird's wings, in my minds eye,
63-
<br>&nbsp;&nbsp; Would be the result of Alpha and Theta waves, influenced by my Frontal Cortex to produce those "meaningless" hypnagogic hallucinations.
88+
<br>&nbsp;&nbsp;&nbsp;&nbsp; Are be the result of Alpha and Theta waves, influenced by my Frontal Cortex to produce those "meaningless" hypnagogic hallucinations.
89+
<br>&nbsp;&nbsp; That would mean there is a distinct line in behaviour between frquencies in sections of the brain.
6490
6591
<br><br><br><div class='procPagesAIDevBar'></div>
6692
6793
<br><br> This reveals a path between a few drastically different areas of my thinking process toward AI development.
6894
6995
<br><br> I'd like to believe it further validates ideas I've been playing with Graph Neural Networks,
70-
<br>&nbsp;&nbsp; As I'm sure I've mentioned the Resonance between neurons in a graph network to spread signals, in prior posts.
96+
<br>&nbsp;&nbsp; As I'm sure I've mentioned in prior posts,
97+
<br>&nbsp;&nbsp;&nbsp;&nbsp; The Resonance between neurons in a graph network to spread signals.
7198
7299
<br><br> That a single neuron vibrating at a certain frequency can harmonize with neurons not directly connected by edges.
73100
<br>&nbsp;&nbsp; The propagation of similarities throughout the local area network of a single neuron.
@@ -78,15 +105,23 @@ entryData.body = `
78105
<br><br> They say we are all 6 degrees from Kevin Bacon,
79106
<br>&nbsp;&nbsp; But in the years to come,
80107
<br>&nbsp;&nbsp; There will have to be a 7th, then 8th, then 9th degree of Kevin Bacon, once he stops being in movies.
81-
<br>&nbsp;&nbsp; And if the rule is, "we are all 6 degrees from Kevin Bacon",
108+
109+
<br><br> If the rule is, "we are all 6 degrees from Kevin Bacon",
82110
<br>&nbsp;&nbsp; The 7th, 8th, 9th would not harmonize with the network of those initial 6.
83111
<br>&nbsp;&nbsp; The Graph Network gets too large,
84-
<br>&nbsp;&nbsp; Harmony is then lost.
112+
<br>&nbsp;&nbsp;&nbsp;&nbsp; Harmony is then lost.
113+
114+
<br><br> This leads me to believe my idea of a tensor landscape affine-projection of the GNN,
115+
<br>&nbsp;&nbsp; Will allow for the harmonization of neurons across the entire network,
116+
<br>&nbsp;&nbsp;&nbsp;&nbsp; Even if some nodes are very far from each other,
117+
<br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Or even disconnected.
85118
86119
<br><br><br><div class='procPagesAIDevBar'></div>
87120
88-
<br><br> Sure, being sleep deprived enough to have two separate hypnagogic hallucinations be about the same red-winged blackbird can't be healthy for me,
89-
<br>&nbsp;&nbsp; But sometimes in our most sleep deprived states,
121+
<br><br> Sure,
122+
<br>&nbsp;&nbsp; Being so sleep deprived to have two separate hypnagogic hallucinations be about the same red-winged blackbird can't be healthy for me,
123+
124+
<br><br> But sometimes in our most sleep deprived states,
90125
<br>&nbsp;&nbsp; The ridiculous becomes the answer we were looking for.
91126
92127
<br><br><div class="textFullRight">- May 5th 2026</div>

0 commit comments

Comments
 (0)