FlowingFlowing
← Back to all development cases
This case was judged by an anonymized AI reviewer and has not been independently reviewed by people; inspect the draft, cards, and both outputs to assess it.
AutocompleteWith cards better76 / 280 · beu-10bc5389b6851aa5

Predictive Coding: a Theoretical and Experimental Review

Psychology and cognitive neuroscience · 2107.12979v4

FLOWING EVIDENCE BENCHMARK

How do we compare writing with and without cards?

At the same writing position, the same model produces one first result with cards and one without. All 280 pairs enter the overall anonymized review, which assesses whether a decision not to continue fits the draft.

Overall: 138 with-card wins, 26 ties, 107 no-card wins, 5 pairs where neither was usable, and 4 inconclusive judgments. Both sides wrote text in 218 pairs; 48 had a deliberate stop and 14 included an invalid output.

SAME DRAFT · WITH OR WITHOUT CARDS

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED VERSIONS

SHARED BY BOTH

Draft, writing task, and generation model

A · With cards

Relevant literature cards supplied

B · Without cards

No literature cards supplied

Generation model

Same model and version

A → first output

B → first output

Anonymized AI judge

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

Review again with the answer order swapped

HOW DOES BLIND JUDGING WORK?

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

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?

We selected 40 writing positions in each of seven fields, for 280 comparisons. The position and relevant literature cards were set before each comparison, so both versions began from the same draft.

The main result retains every frozen position, including places where cards might not add a useful next move. This avoids selecting cases based on their outputs, but 49% describes the whole cohort, not a win rate where evidence clearly had an opportunity to help. This cohort did not record that opportunity label for each position before generation, so those two groups cannot yet be separated reliably.

CARD USE · 140/280

Reported card use in continuations with text

The model reports whether a card affected its continuation. In 140 of 280 tests it produced continuation text and reported card use. The previous count of 150 included 10 cases without continuation text and has been corrected.

COMPARISON · 70 / 7 / 30

Which written version was better?

Of the 107 reported-use cases where both versions produced text, the blinded AI judge preferred the with-card version 70 times, found 7 about the same, and preferred the without-card version 30 times.

How are the two table percentages calculated?

ADOPTION COVERAGE

140reported use with text
÷
280all positions
=
50%

WITH-CARD VERSION PREFERRED

70with cards better
÷
107comparable tests
=
65%

The percentages use different groups: 50% counts continuations with text reporting card use among all 280 tests; 65% covers the 107 such cases with text on both sides. The generating model reports card use, while an anonymized AI judge compares writing quality.

Card use is self-reported by the generator. The anonymized AI judge sees the draft and source snippets, so source-aware preference remains possible; there was no independent human review. The overall result describes these 280 frozen positions, not a win probability for any writing task.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

Predictive Coding: a Theoretical and Experimental Review

5 Discussion and Future Directions

… ive-adversarial ( Goodfellow et al., (2014) ) and variational autoencoding models ( Kingma & Welling, (2013) ; Child, (2020) ) , as well as recent work showing the considerable power of unsupervised contrastive objectives to learn powerful representations in audio ( Oord et al., (2016) ) , and text and images ( Radford et al., (2021) ) . The question now remains, is there anything additional that predictive coding can bring to machine learning? Perhaps the key idea that predictive coding possesses, which is lacking in current machine learning, is a strong role for recurrent, top-down connectivity. In many ways, the success of purely feedforward machine learning architectures (which even in the case of transformers outperform previously recurrent architectures such as the LSTM in natural language tasks) has retroactively bolstered much of the ‘classical’ view of visual perception as a pure feedforward pass through a set of hierarchically composable feature detectors. This approach is precisely the one taken by large scale machine learning models, and has been shown to be able to achieve human-level object recognition and detection in images ( Krizhevsky et al., (2012) ) , as well as matching to a surprising degree the observed activations in biological brains ( Yamins et al., (2014) ) . Nevertheless, it is possible that performance may be further improved, and the models themselves may be made more flexible by the introduction of a recurrent top-down processing stream. There are many theoretical advantages of recurrence ( van Bergen & Kriegeskorte, (2020) ; Kietzmann et al.,

↓ 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 positionOverall AI review: With cards better

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.

Read overall AI review rationale

A = with cards · B = without cards

Both outputs attempt to continue the sentence after 'Kietzmann et al.,' which is clearly an incomplete citation. The draft is making an argument about theoretical advantages of recurrence and needs to develop this point. Output B simply completes the citation parenthetical with '(2020)' and adds a brief clause about noise/occlusion robustness and iterative refinement. This is grammatically awkward because it stuffs content inside the citation parenthetical rather than closing it first, and the content is somewhat generic. Output A closes the citation (implicitly, by moving to a new sentence) and provides a substantive continuation that connects back to the broader argument: it brings in PredNet as a concrete example of how predictive coding's recurrence/top-down structure can support unsupervised learning, directly addressing the paper's central question about what predictive coding can bring to machine learning. Output A better fits the argumentative arc, has appropriate scope, and could be inserted with only minor grammatical adjustment (adding the citation year). Output B's grammatical structure is more problematic and its contribution more superficial.

Evidence

Same model and position · retrieved passages supplied · first output

Preferred by blind review

In particular, predictive coding architectures such as PredNet demonstrate that prediction error propagation can support unsupervised learning of object and scene structure from video, suggesting a concrete route by which top-down recurrence might augment feedforward representation learning.

Passages supplied to the Evidence version

Card 1 · Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning

While great strides have been made in using deep learning algorithms to solve supervised learning tasks, the problem of unsupervised learning - leveraging unlabeled examples to learn about the structure of a domain - remains a difficult unsolved challenge. Here, we explore prediction of future frame…

Read full passage excerpt

While great strides have been made in using deep learning algorithms to solve supervised learning tasks, the problem of unsupervised learning - leveraging unlabeled examples to learn about the structure of a domain - remains a difficult unsolved challenge. Here, we explore prediction of future frames in a video sequence as an unsupervised learning rule for learning about the structure of the visual world. We describe a predictive neural network ("PredNet") architecture that is inspired by the concept of "predictive coding" from the neuroscience literature. These networks learn to predict future frames in a video sequence, with each layer in the network making local predictions and only forwarding deviations from those predictions to subsequent network layers. We show that these networks are able to robustly learn to predict the movement of synthetic (rendered) objects, and that in doing so, the networks learn internal representations that are useful for decoding latent object parameter …

Card 2 · PredNet and Predictive Coding: A Critical Review

PredNet, a deep predictive coding network developed by Lotter et al., combines a biologically inspired architecture based on the propagation of prediction error with self-supervised representation learning in video. While the architecture has drawn a lot of attention and various extensions of the mo…

Read full passage excerpt

PredNet, a deep predictive coding network developed by Lotter et al., combines a biologically inspired architecture based on the propagation of prediction error with self-supervised representation learning in video. While the architecture has drawn a lot of attention and various extensions of the model exist, there is a lack of a critical analysis. We fill in the gap by evaluating PredNet both as an implementation of the predictive coding theory and as a self-supervised video prediction model using a challenging video action classification dataset. We design an extended model to test if conditioning future frame predictions on the action class of the video improves the model performance. We show that PredNet does not yet completely follow the principles of predictive coding. The proposed top-down conditioning leads to a performance gain on synthetic data, but does not scale up to the more complex real-world action classification dataset. Our analysis is aimed at guiding future research …

Card 3 · Predictive Coding Can Do Exact Backpropagation on Convolutional and Recurrent Neural Networks

Predictive coding networks (PCNs) are an influential model for information processing in the brain. They have appealing theoretical interpretations and offer a single mechanism that accounts for diverse perceptual phenomena of the brain. On the other hand, backpropagation ( BP ) is commonly regarded…

Read full passage excerpt

Predictive coding networks (PCNs) are an influential model for information processing in the brain. They have appealing theoretical interpretations and offer a single mechanism that accounts for diverse perceptual phenomena of the brain. On the other hand, backpropagation ( BP ) is commonly regarded to be the most successful learning method in modern machine learning. Thus, it is exciting that recent work formulates inference learning ( IL ) that trains PCNs to approximate BP. However, there are several remaining critical issues: (i) IL is an approximation to BP with unrealistic/non-trivial requirements, (ii) IL approximates BP in single-step weight updates; whether it leads to the same point as BP after the weight updates are conducted for more steps is unknown, and (iii) IL is computationally significantly more costly than BP.

Card 4 · PreCNet: Next-Frame Video Prediction Based on Predictive Coding

Predictive coding, currently a highly influential theory in neuroscience, has not been widely adopted in machine learning yet. In this work, we transform the seminal model of Rao and Ballard (1999) into a modern deep learning framework while remaining maximally faithful to the original schema. The r…

Read full passage excerpt

Predictive coding, currently a highly influential theory in neuroscience, has not been widely adopted in machine learning yet. In this work, we transform the seminal model of Rao and Ballard (1999) into a modern deep learning framework while remaining maximally faithful to the original schema. The resulting network we propose (PreCNet) is tested on a widely used next-frame video prediction benchmark, which consists of images from an urban environment recorded from a car-mounted camera, and achieves state-of-the-art performance. Performance on all measures (MSE, PSNR, and SSIM) was further improved when a larger training set (2M images from BDD100k) pointed to the limitations of the KITTI training set. This work demonstrates that an architecture carefully based on a neuroscience model, without being explicitly tailored to the task at hand, can exhibit exceptional performance.

Card 5 · Deep Predictive Coding Networks

The quality of data representation in deep learning methods is directly related to the prior model imposed on the representations; however, generally used fixed priors are not capable of adjusting to the context in the data. To address this issue, we propose deep predictive coding networks, a hierar…

Read full passage excerpt

The quality of data representation in deep learning methods is directly related to the prior model imposed on the representations; however, generally used fixed priors are not capable of adjusting to the context in the data. To address this issue, we propose deep predictive coding networks, a hierarchical generative model that empirically alters priors on the latent representations in a dynamic and context-sensitive manner. This model captures the temporal dependencies in time-varying signals and uses top-down information to modulate the representation in lower layers. The centerpiece of our model is a novel procedure to infer sparse states of a dynamic model which is used for feature extraction. We also extend this feature extraction block to introduce a pooling function that captures locally invariant representations. When applied on a natural video data, we show that our method is able to learn high-level visual features. We also demonstrate the role of the top-down connections by s …

Model-reported card use

Card 1 · fact

Affected output: predictive coding architectures such as PredNet demonstrate that prediction error propagation can support unsupervised learning of object and scene structure from video

Model explanation: This row supplied the concrete example of PredNet and its prediction-error propagation mechanism, which grounds the claim that predictive coding architectures can support unsupervised learning from video.

No Evidence

Same model and position · no retrieved passages · first output

Not preferred in this comparison

2020), such as improved robustness to noise and occlusion, and the ability to iteratively refine predictions over time.

Earlier card-use comparison

The earlier diagnostic compared only cases with reported card use and text on both sides; its scope differs from the overall AI review above.

The blinded model review preferred the no-Evidence version. Its original rationale appears below.

Read the original model judgment

Output B provides a grammatically correct continuation that completes the interrupted citation '( Kietzmann et al.,' and lists concrete theoretical advantages of recurrence (robustness to noise/occlusion, iterative refinement) that are well-established in the literature and consistent with the draft's argument about top-down processing. The year '2020)' properly closes the citation, and the content flows naturally from 'There are many theoretical advantages of recurrence.' Output A, while topically relevant to predictive coding, introduces an unsupported claim: it states that 'PredNet demonstrate[s] that prediction error propagation can support unsupervised learning of object and scene structure from video' and that this 'suggest[s] a concrete route by which top-down recurrence might augment feedforward representation learning.' The sources provided do not establish that PredNet demonstrates top-down recurrence augmenting feedforward representation learning. Source 1 (PredNet paper) discusses prediction and unsupervised learning but does not claim top-down recurrence augments feedforward learning. Source 2 (critical review) explicitly states 'PredNet does not yet completely follow the principles of predictive coding' and that top-down conditioning 'does not scale up.' Source 4 (PreCNet) achieves state-of-the-art but does not claim top-down recurrence augments feedforward learning. The claim in A about 'concrete route by which top-down recurrence might augment feedforward representation learning' is an extrapolation not supported by the sources. Additionally, A ignores the incomplete citation that needs completion. B correctly handles the citation and provides grounded, general theoretical advantages without overclaiming specific source support.

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.