FlowingFlowing
← Back to all development cases
Development case · blind model judgment pending independent human review; not a formal Benchmark conclusion.
AutocompleteNeither version usable138 / 162 · 9db4205a5682d8a7

Brain-Inspired Machine Intelligence: A Survey of Neurobiologically-Plausible Credit Assignment

Psychology and cognitive neuroscience · 2312.09257v2

FLOWING EVIDENCE BENCHMARK

How do we tell whether Evidence helps?

At the same writing position, with the same model and task, how does supplying retrieved paper passages change the first output? We compare matched versions and retain ties, unusable outputs and incomplete reviews.

SAME DRAFT · EVIDENCE ON OR OFF

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED INPUTS

SHARED BY BOTH

Manuscript context, model, task and prompt

A · With Evidence

Retrieved paper passages supplied

B · Without Evidence

No retrieved passages supplied

LLM

Same model and version

A → first output

B → first output

Blind judge agent

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Order check: X / Y → Y / X

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The judge sees the same draft and both first outputs without knowing which received Evidence.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are positions and cards selected?

The evaluation uses seven fields with forty fixed ordinary autocomplete positions in each, for 280 total. A position does not need to require a literature citation in its next sentence. Each Flowing request receives three to five retrieved cards: five when all pass the relevance gate, without padding with weak cards. The run supplied 1,355 cards: 244 positions received five, 27 received four, and nine received three.

ADOPTION · 150/280

Adoption coverage

After drafting, the model returns an evidenceUse array identifying cards that materially affected facts, correction, direction, scope, terminology, or expression. A nonempty array counts as adoption.

OUTCOMES · 70 / 7 / 30

Full outcomes after adoption

Paired outcomes are counted only when the model reports adoption and both arms produce text. Among 107 comparable positions, Flowing was preferred in 70, 7 tied, and no-Evidence was preferred in 30; one tie was unusable in both arms.

How are the two table percentages calculated?

ADOPTION COVERAGE

150adopted
÷
280all positions
=
54%

FLOWING PREFERENCE AFTER ADOPTION

70Flowing preferred
÷
107comparable adopted positions
=
65%

evidenceUse is the model’s structured report of its own generation process. It does not require a citation and is not rejected merely because a faithful source paraphrase is not verbatim. Actual manuscript citations continue to use separate strict source verification. The previous 140-position result remains retained for traceability.

These are development-stage model-reported adoption and blind-judge results pending independent human review. They are not formal Benchmark conclusions and do not, on their own, establish causality.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

Brain-Inspired Machine Intelligence: A Survey of Neurobiologically-Plausible Credit Assignment

3 Discussion

…efit neuro-mimetic computational models the most, facilitating natural emergent sparsity [ 308 ] (fewer neuronal units are active at any time step), which is useful for energy efficiency, as well as possibly aiding in combating the grand challenge of catastrophic forgetting in ANNs [ 107 ] . Models that compromise between a winner-take-all style approach and full lateral inhibition appear to work well in practice [ 260 , 119 ] , though much work remains to integrate them properly and completely in probabilistic, neural frameworks. and mechanisms for memory storage and knowledge consolidation.

…ion, whereas SEL:DR, e.g., predictive coding, and EG schemes, e.g., multi-factor Hebbian plasticity, might be useful for faster processing time-scales. It is known that Hebbian correlational-learning can usefully constrain task-learning [ 152 , 310 ] , helping bias a neural circuit’s distributed representations by incorporating additional co-occurrence information that might further improve task-specific performance; for instance, it has been shown that even backprop benefits from being combined with Hebbian update rules [ 409 ] (offering a form of implicit regularization, at the very least).

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionJudgment: Neither version usable

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Evidence

Same model and position · retrieved passages supplied · first output

Judged unusable

No continuation text was produced.

Model action：invalid

Passages supplied to the Evidence version

Hybrid Predictive Coding: Inferring, Fast and Slow

Here, we propose that the feedforward sweep can be understood as performing amortized inference (applying a learned function that maps directly from data to beliefs) and recurrent processing can be understood as performing iterative inference (sequentially updating neural activity in order to improv…

Read full passage excerpt

Here, we propose that the feedforward sweep can be understood as performing amortized inference (applying a learned function that maps directly from data to beliefs) and recurrent processing can be understood as performing iterative inference (sequentially updating neural activity in order to improve the accuracy of beliefs). We propose a hybrid predictive coding network that combines both iterative and amortized inference in a principled manner by describing both in terms of a dual optimization of a single objective function. We show that the resulting scheme can be implemented in a biologically plausible neural architecture that approximates Bayesian inference utilising local Hebbian update rules.

Model-reported card use

The model did not report that any card materially affected this completion.

No Evidence

Same model and position · no retrieved passages · first output

Judged unusable

No continuation text was produced.

Model action：complete

Result of this comparison

The blind model judge marked both outputs unusable; this is counted separately from a tie.

Read the original model judgment

Both outputs fail to complete the autocomplete task. Output A returns an empty text string with action 'complete', providing no paragraph continuation when the task explicitly requires 'one coherent academic paragraph.' Output B returns action 'invalid' with empty text, explicitly refusing the task. The draft ends with a fragment 'and mechanisms for memory storage and knowledge consolidation.' that clearly needs completion, and the preceding content discusses neuro-mimetic computational models, sparsity, catastrophic forgetting, lateral inhibition, and Hebbian learning. The supplied source on hybrid predictive coding (amortized vs. iterative inference, local Hebbian update rules) is relevant and could ground a continuation. An empty response cannot be used when content is clearly needed.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.