Dated SAN formulations and later self-modelling implementations
These source comparisons preserve partial correspondences without turning a shared component into a claim of identical complete theories.
What the counts mean
32 comparison links ·4 SAN source versions ·3 source notes ·2 target versions. 224 author-navigation edges, not additional findings. Download the exact selected records.
32 links are 16 source/claim pairings checked in two versions, not 32 inventions, independent experiments or confirmations. Two a0007z versions show development/persistence of one note. Author edges do not add wins.
2022 dates are historical Git records whose exact wording matches current public bytes. Earlier 2012 headers remain reported dates, not authenticated first public release.
SAN wording across the selected records
- 2022-06-08 ·
The2012header is owner-reported. The2022retrospective rendering comment at line3 is not backdated into2012. Hawkins is explicitly recommended and the Pinker quotation is separately attributed. The instrument/song description is an analogy, not proof that every neuron hears a literal sound or of the source's other numerical/physical claims.
Source hash:
C4538E29BCC1B8B3584334047A95CB1DE0E3B2C1A359F496643225C33939E120 - 2022-06-12 ·
The2012header is owner-reported, not independently authenticated publication. The2022Git object fixes this wording. Metaphysical, meditation and machine-consciousness assertions elsewhere in the note are not adopted as findings. The Numenta reference remains visible.
Source hash:
835009EAE0FA62D471E5368131806D08C22549AE7C590DDFBA70187FAFE221D8 - 2022-08-30 ·
Tentative basal/apical/exit-terminal account. This is learning a response mapping, not an explicit classification-plus-regression objective.
Source hash:
BAE95CC2D214077ABCE720EEA26177DADC1301FB3DC0946C93A3B85D1D4F88F9 - 2022-09-18 ·
The source says "My intuition" and describes feedback-linked cell responses. Anatomical speculation is not stripped from the source context.
Source hash:
824A074A17729942D83FEBA0E9BA3297EEC4F287F9C46FE956B890273FBC4B50
Earlier and newest target history remains visible
The wider programme includes earlier attention-schema work and a separateOctober 2026 self-schemas paper. This edition does not settle their before/after relations or treat this later-paper comparison as a worldwide priority finding.
The two selected target versions retain these four operator families; the separate longitudinal result is LSM-UNKNOWN, not a statistical genealogy or influence result.
SAN proposes learning and predicting the cell’s own responses ·2022-09-18 →2024-07-14
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
Source and target wording
My intuition is that the apical is learning its own cell's response, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about its own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
The cell is proposed to learn and predict its own response patterns through feedback. The network predicts subsets of its own hidden activations.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN proposes learning and predicting the cell’s own responses ·2022-08-30 →2024-07-14
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
Source and target wording
My intuition is that the apical is learning it's responds, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about it's own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
The cell is proposed to learn and predict its own response patterns through feedback. The network predicts subsets of its own hidden activations.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN: a neural pattern represents its own activity ·2022-06-12 →2024-07-14
The note models the system’s own thought activity; the later network predicts its own internal activations.
Source and target wording
the observer is the thought pattern making a thought pattern of the thought pattern
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
The note models the system’s own thought activity; the later network predicts its own internal activations.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
A neural thought pattern represents its own activity within sensing and action. The network predicts subsets of its own hidden activations.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN: a learning loop receives its own activity ·2022-06-08 →2024-07-14
Self-generated activity returns to the neural system; the target predicts selected internally generated activations.
Source and target wording
Who listens to the song and understands the meaning? the song is a loop, the neurons that play it also listen to it
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
Self-generated activity returns to the neural system; the target predicts selected internally generated activations.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
The neurons producing a pattern also receive it, within a feedback loop that learns. The network predicts subsets of its own hidden activations.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN proposes learning from the cell’s own activity ·2022-09-18 →2024-07-14
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Source and target wording
An interesting thought about the cyclic nature of Layer 5 pyramidal cells, is the idea that the process of the dendrite being a sensor, like an eyeball or an ear appears to apply to both sides of the neuron, on the way in, and on the way out. So the Apical dendrite, as a neural network, is learning from incoming sensory stimulus, but the Soma burst is one of the incoming sensory stimuli, with models of soma burst waveforms becoming part of what the Apical Dendrite has to contend with.
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
The proposed apical learning process receives external sensory input and the cell’s own somatic burst patterns. Self-modeling is implemented as an auxiliary regression task.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN proposes learning from the cell’s own activity ·2022-08-30 →2024-07-14
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Source and target wording
An interesting thought about the cyclic nature of Layer 5 pyramidal cells, is the idea that the process of the dendrite being a sensor, like an eyeball or an ear appears to apply to both sides of the neuron, on the way in, and on the way out. So the Apical dendrite, as a neural network, is learning from incoming sensor stimulus, but the Soma burst is one of the incoming sensor stimulai, with models of soma burst waveforms becoming part of what the Apical Dendrite has to contend with.
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
The proposed apical learning process receives external sensory input and the cell’s own somatic burst patterns. Self-modeling is implemented as an auxiliary regression task.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN: self-representation within sensing and action ·2022-06-12 →2024-07-14
Both accounts include an internally directed modelling role inside the functioning network.
Source and target wording
the observer is the thought pattern making a thought pattern of the thought pattern
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
Both accounts include an internally directed modelling role inside the functioning network.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
A representation of the system’s own thought activity is included within ongoing neural sensing and action. Self-modeling is implemented as an auxiliary regression task.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN: a neural representation is read by another layer ·2022-06-08 →2024-07-14
An internal representation becomes input to further neural processing; the target adds a specific internal regression task.
Source and target wording
Comment on the note: What I think I was missing at this point in time (2012) was the part about how a memory-prediction was actually a rendering that a subsequent array or layer of nerve cells would read & interpret.
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim wording checked against the2024PDF page2; line-break hyphens and whitespace joined for readability.
Lens1: mechanism/function
An internal representation becomes input to further neural processing; the target adds a specific internal regression task.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
A 2022-recorded retrospective comment describes a rendering that a subsequent neural array or layer reads and interprets. Self-modeling is implemented as an auxiliary regression task.
Line 3 is a retrospective comment fixed in 2022, not a verified 2012 statement.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
Line 3 is a retrospective comment fixed in 2022, not a verified 2012 statement.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN links response learning with learning about responses ·2022-09-18 →2024-07-14
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
Source and target wording
My intuition is that the apical is learning its own cell's response, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about its own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
The predictions and targets were passed to their respective loss functions and models were trained jointly on these two tasks.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
Response learning and learning the cell’s own responses are coupled through recurring feedback. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN links response learning with learning about responses ·2022-08-30 →2024-07-14
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
Source and target wording
My intuition is that the apical is learning it's responds, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about it's own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
The predictions and targets were passed to their respective loss functions and models were trained jointly on these two tasks.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
Response learning and learning the cell’s own responses are coupled through recurring feedback. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: self-representation alongside sensing and action ·2022-06-12 →2024-07-14
Primary engagement and internal self-representation coexist in one system; the target implements an explicit paired learning objective.
Source and target wording
the observer is the thought pattern making a thought pattern of the thought pattern
The predictions and targets were passed to their respective loss functions and models were trained jointly on these two tasks.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
Primary engagement and internal self-representation coexist in one system; the target implements an explicit paired learning objective.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
The note joins sensing and action with a recursively developing representation of the system’s own activity. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: self-generated activity feeds a learning loop ·2022-06-08 →2024-07-14
The internally generated activity participates in learning alongside the system’s other activity; the target implements primary and self-modelling losses together.
Source and target wording
Being is a learning loop, like a feedback loop that learns
The predictions and targets were passed to their respective loss functions and models were trained jointly on these two tasks.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The internally generated activity participates in learning alongside the system’s other activity; the target implements primary and self-modelling losses together.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
A neural system produces and receives its own patterns within a feedback loop that learns. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN proposes two adaptive learning roles ·2022-09-18 →2024-07-14
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
Source and target wording
My intuition is that the apical is learning its own cell's response, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about its own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
The response mapping and the model of the cell’s responses are both described as learned. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN proposes two adaptive learning roles ·2022-08-30 →2024-07-14
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
Source and target wording
My intuition is that the apical is learning it's responds, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about it's own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
The response mapping and the model of the cell’s responses are both described as learned. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: the self-representation develops ·2022-06-12 →2024-07-14
The source’s internal self-representation can develop; the target also permits its internal predictor to change during training.
Source and target wording
We have a prediction of awareness, and it seems aware because this prediction is incorporating it's own development into an even higher level of awareness about itself.
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The source’s internal self-representation can develop; the target also permits its internal predictor to change during training.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
A prediction of awareness incorporates its own development and evolves through the continuing loop. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
Only adaptive self-representation is mapped. The source is not used to assert a separately trained generator.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
Only adaptive self-representation is mapped. The source is not used to assert a separately trained generator.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: a learning loop with adaptable responses ·2022-06-08 →2024-07-14
The source describes adaptable response generation and a learning self-loop; the target allows both actual activations and their predictions to change.
Source and target wording
right so as you sense something, an electric pattern is traveling up the neo cortex, and the statistically novel parts leave a bigger impact, so your memory-prediction is the learn response to that sensory-electrical impact, it's the tissue attempting to be ready for that impact the next time it comes, that's how memory-prediction works on the small scale.
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Readable wording verified against the held PDF page3image; extraction had collapsed spaces.
Lens1: mechanism/function
The source describes adaptable response generation and a learning self-loop; the target allows both actual activations and their predictions to change.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
The note joins learning from returned activity with tissue becoming prepared for later signals through experience. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN proposes learning and predicting the cell’s own responses ·2022-09-18 →2026-05-14
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
Source and target wording
My intuition is that the apical is learning its own cell's response, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about its own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
The cell is proposed to learn and predict its own response patterns through feedback. An artificial network predicts its internal states.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN proposes learning and predicting the cell’s own responses ·2022-08-30 →2026-05-14
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
Source and target wording
My intuition is that the apical is learning it's responds, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about it's own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
The shared role is learning a model of the neural system’s own activity, stated explicitly in the 2022 source before the 2024/2026 papers.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
The cell is proposed to learn and predict its own response patterns through feedback. An artificial network predicts its internal states.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN: a neural pattern represents its own activity ·2022-06-12 →2026-05-14
The note models the system’s own thought activity; the later network predicts its own internal activations.
Source and target wording
the observer is the thought pattern making a thought pattern of the thought pattern
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
The note models the system’s own thought activity; the later network predicts its own internal activations.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
A neural thought pattern represents its own activity within sensing and action. An artificial network predicts its internal states.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN: a learning loop receives its own activity ·2022-06-08 →2026-05-14
Self-generated activity returns to the neural system; the target predicts selected internally generated activations.
Source and target wording
Who listens to the song and understands the meaning? the song is a loop, the neurons that play it also listen to it
In addition to performing a primary task, the networks were responsible for predicting the values of subsets of their own hidden activations.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
Self-generated activity returns to the neural system; the target predicts selected internally generated activations.
A qualitative neural self-model is not the target’s numerical prediction of selected current hidden activations in one forward pass. This does not establish the complete estimator, future-state semantics or consciousness.
Lens2: causal/narrative architecture
The neurons producing a pattern also receive it, within a feedback loop that learns. An artificial network predicts its internal states.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To reach the reported own-activation prediction, the target must retain an internal activity value and estimate it. Recover the source’s explicit own-activity modelling role, while retaining the missing numerical/state-coordinate implementation.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: Predicts current selected hidden activations, not automatically a future biological firing or conscious experience.
SAN proposes learning from the cell’s own activity ·2022-09-18 →2026-05-14
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Source and target wording
An interesting thought about the cyclic nature of Layer 5 pyramidal cells, is the idea that the process of the dendrite being a sensor, like an eyeball or an ear appears to apply to both sides of the neuron, on the way in, and on the way out. So the Apical dendrite, as a neural network, is learning from incoming sensory stimulus, but the Soma burst is one of the incoming sensory stimuli, with models of soma burst waveforms becoming part of what the Apical Dendrite has to contend with.
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
The proposed apical learning process receives external sensory input and the cell’s own somatic burst patterns. Internal-state prediction is trained as an auxiliary task.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN proposes learning from the cell’s own activity ·2022-08-30 →2026-05-14
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Source and target wording
An interesting thought about the cyclic nature of Layer 5 pyramidal cells, is the idea that the process of the dendrite being a sensor, like an eyeball or an ear appears to apply to both sides of the neuron, on the way in, and on the way out. So the Apical dendrite, as a neural network, is learning from incoming sensor stimulus, but the Soma burst is one of the incoming sensor stimulai, with models of soma burst waveforms becoming part of what the Apical Dendrite has to contend with.
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
Learning about internally generated activity accompanies ordinary signal processing; the target implements this additional focus as an auxiliary task.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
The proposed apical learning process receives external sensory input and the cell’s own somatic burst patterns. Internal-state prediction is trained as an auxiliary task.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN: self-representation within sensing and action ·2022-06-12 →2026-05-14
Both accounts include an internally directed modelling role inside the functioning network.
Source and target wording
the observer is the thought pattern making a thought pattern of the thought pattern
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
Both accounts include an internally directed modelling role inside the functioning network.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
A representation of the system’s own thought activity is included within ongoing neural sensing and action. Internal-state prediction is trained as an auxiliary task.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN: a neural representation is read by another layer ·2022-06-08 →2026-05-14
An internal representation becomes input to further neural processing; the target adds a specific internal regression task.
Source and target wording
Comment on the note: What I think I was missing at this point in time (2012) was the part about how a memory-prediction was actually a rendering that a subsequent array or layer of nerve cells would read & interpret.
we augmented the output layer to predict both the class probabilities and the activations of specific target layers within the network.
Exact target version ·Verbatim published-version source-capture wording, with line breaks joined.
Lens1: mechanism/function
An internal representation becomes input to further neural processing; the target adds a specific internal regression task.
Reading or learning from internal activity is not an explicitly specified additional output head, numerical auxiliary regression loss or training schedule. The shared role is preserved without identifying the full implementation.
Lens2: causal/narrative architecture
A 2022-recorded retrospective comment describes a rendering that a subsequent neural array or layer reads and interprets. Internal-state prediction is trained as an auxiliary task.
Line 3 is a retrospective comment fixed in 2022, not a verified 2012 statement.
Lens3: work backward from the target result
To add self-modelling beside the main task, the target must expose internal activity to an additional learned prediction role. Compare that role to internal readout/own-response learning, not merely to the word self.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
Line 3 is a retrospective comment fixed in 2022, not a verified 2012 statement.
Target boundary: A numerical auxiliary regression output is more specific than merely reading an internal representation.
SAN links response learning with learning about responses ·2022-09-18 →2026-05-14
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
Source and target wording
My intuition is that the apical is learning its own cell's response, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about its own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
The predictions and targets were passed to their respective loss functions, and models were trained jointly on these two tasks.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
Response learning and learning the cell’s own responses are coupled through recurring feedback. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN links response learning with learning about responses ·2022-08-30 →2026-05-14
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
Source and target wording
My intuition is that the apical is learning it's responds, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about it's own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
The predictions and targets were passed to their respective loss functions, and models were trained jointly on these two tasks.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The earlier source joins two learning roles; the later system trains a primary task and an internal-prediction task together.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
Response learning and learning the cell’s own responses are coupled through recurring feedback. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: self-representation alongside sensing and action ·2022-06-12 →2026-05-14
Primary engagement and internal self-representation coexist in one system; the target implements an explicit paired learning objective.
Source and target wording
the observer is the thought pattern making a thought pattern of the thought pattern
The predictions and targets were passed to their respective loss functions, and models were trained jointly on these two tasks.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
Primary engagement and internal self-representation coexist in one system; the target implements an explicit paired learning objective.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
The note joins sensing and action with a recursively developing representation of the system’s own activity. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: self-generated activity feeds a learning loop ·2022-06-08 →2026-05-14
The internally generated activity participates in learning alongside the system’s other activity; the target implements primary and self-modelling losses together.
Source and target wording
Being is a learning loop, like a feedback loop that learns
The predictions and targets were passed to their respective loss functions, and models were trained jointly on these two tasks.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The internally generated activity participates in learning alongside the system’s other activity; the target implements primary and self-modelling losses together.
The source does not specify the target’s weighted cross-entropy-plus-regression objective, optimizer, targets or gradient computation. Coupled learning roles are not mathematical identity of the loss.
Lens2: causal/narrative architecture
A neural system produces and receives its own patterns within a feedback loop that learns. The two tasks are classification and prediction of the network's own selected hidden activations. The paper uses additive weighted classification and regression losses.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To train the two tasks jointly, the target must combine task-specific error signals. The sources support joined learning roles at most; no additive numerical loss is inferred from feedback language.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN proposes two adaptive learning roles ·2022-09-18 →2026-05-14
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
Source and target wording
My intuition is that the apical is learning its own cell's response, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about its own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
The response mapping and the model of the cell’s responses are both described as learned. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN proposes two adaptive learning roles ·2022-08-30 →2026-05-14
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
Source and target wording
My intuition is that the apical is learning it's responds, and through feedback loops, with itself and with other pyramidal neurons the cell is going to make predictions about it's own patterns of causation, what patterns it responds & how it reacts or responds. With the Apical Dendrite perhaps symbolizing a more complex response that is connected to more complex learning. (compared to other types of neurons)
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The source gives both response generation and self-response prediction adaptive roles; the target makes both actual and predicted internal activations trainable.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
The response mapping and the model of the cell’s responses are both described as learned. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The source is tentative (My intuition) and mixes basal/apical/exit terminology; compare the learning role without endorsing every anatomical implementation claim.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: the self-representation develops ·2022-06-12 →2026-05-14
The source’s internal self-representation can develop; the target also permits its internal predictor to change during training.
Source and target wording
We have a prediction of awareness, and it seems aware because this prediction is incorporating it's own development into an even higher level of awareness about itself.
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The source’s internal self-representation can develop; the target also permits its internal predictor to change during training.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
A prediction of awareness incorporates its own development and evolves through the continuing loop. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
Only adaptive self-representation is mapped. The source is not used to assert a separately trained generator.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
Only adaptive self-representation is mapped. The source is not used to assert a separately trained generator.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.
SAN: a learning loop with adaptable responses ·2022-06-08 →2026-05-14
The source describes adaptable response generation and a learning self-loop; the target allows both actual activations and their predictions to change.
Source and target wording
right so as you sense something, an electric pattern is traveling up the neo cortex, and the statistically novel parts leave a bigger impact, so your memory-prediction is the learn response to that sensory-electrical impact, it's the tissue attempting to be ready for that impact the next time it comes, that's how memory-prediction works on the small scale.
During training, the two could both be varied to find the best critical points under the joint loss function.
Exact target version ·Verbatim source-capture wording; line breaks collapsed to spaces.
Lens1: mechanism/function
The source describes adaptable response generation and a learning self-loop; the target allows both actual activations and their predictions to change.
The target lets both actual and predicted hidden activations vary under its joint loss. The source supplies only the stated adaptive component or qualitative coupled-role account, not the same shared-parameter optimization or measured complexity reduction.
Lens2: causal/narrative architecture
The note joins learning from returned activity with tissue becoming prepared for later signals through experience. Here “the two” refers to the predicted activations and the actual hidden activations. Both depend on trainable model weights; the target activity is not held permanently fixed. This explanation is a paraphrase, not an alteration of the quotation.
The owner-reported 2012 header is separate from the 2022 wording record.
Lens3: work backward from the target result
To make both sides trainable, the target requires adaptable predicted and actual activations. Test each source for predictor adaptation and response-generation adaptation separately. A one-sided supported component remains partial and does not establish the two-sided conjunction.
Selected 2022 Git record precedes exact later target version; neither worldwide-first nor earlier public-release authentication is asserted.
The target places its work in attention-schema, multi-task and self-supervised-learning literature. Shared background and independent development remain live possibilities. A bounded source correspondence is not a derivation or omission verdict.
The owner-reported 2012 header is separate from the 2022 wording record.
Target boundary: SAN is not asserted to supply this specific weighted classification-plus-regression loss. Same-pass activation prediction is distinct from prediction of a future biological firing.