Meeting Notes

• Date: 2026-02-10
• Time: 09:00AM (PT)
• Location: Teams Meeting

Agenda

Discussion on figure update

• @jeromelecoq will present the updated figure for the graphical abstract
• @lrudelt will present the updated figure 2.

We can answer questions on all of those.

Meeting Recording

Meeting Notes

Graphical Abstract Review and Feedback: Jerome led a discussion with Benedicte, Stefan, and others on the graphical abstract for the review paper, focusing on its clarity, the rationale for design choices such as the use of a mouse image, and the representation of hierarchical scales, with Stefan and Martin suggesting modifications to the use of arrows and box zooms to avoid implying a hierarchy of importance.

Design Rationale: Jerome explained that the graphical abstract was designed to summarize predictive processing mechanisms at different scales, starting with stimulus design and moving through brain-wide, local, and single-cell computations, with a focus on the mouse due to the abundance of data at all levels.

Arrow and Box Representation Debate: Stefan questioned the use of large arrows between boxes, suggesting they implied a hierarchy of importance, and proposed using a zoom-in box or lens approach instead; Martin agreed, recommending the removal of arrows to avoid suggesting unnecessary hierarchy.

Consensus on Figure Adjustments: The group, including Benedicte and Stefan, reached consensus to remove most arrows for clarity, possibly retaining the first arrow to indicate the transition from stimulus to mouse, and to rely on the zoomed-in box structure to convey scale transitions.

Clarity of Stimulus Box: Jerome raised the issue of clarity regarding the box inside the stimulus, referencing prior confusion and edits, and confirmed with the group that the current version was now clear to all present.

Figure 2 Revision: Hierarchical and Non-Hierarchical Predictive Coding: Lucas, with input from Jerome, John, Stefan, Karim, Benedicte, and Martin, presented alternative versions of Figure 2 depicting hierarchical and non-hierarchical implementations of predictive coding, leading to an in-depth discussion on circuit representation, color coding, and the depiction of inter-area connections, with the group agreeing to further refine the figure and consider additional schematic proposals.

Presentation of Figure Versions: Lucas presented both the previous and a new version of Figure 2, explaining the depiction of classical hierarchical predictive coding and alternative proposals involving positive and negative error neurons, referencing influential literature and recent drafts.

Debate on Non-Hierarchical Implementation: Karim questioned the output of the non-hierarchical module, noting ambiguity in the original literature, and advocated for clarity in the figure to avoid reviewer confusion; Lucas and Jerome acknowledged the need to fill in unspecified aspects for completeness.

Preference for Hierarchical Representation: John and Stefan argued for retaining the hierarchical implementation in the figure, emphasizing the importance of depicting input and output areas and warning against showing modules without clear outputs, while Lucas agreed to focus on the version consistent with hierarchical structure.

Color Coding and Circuit Consistency: Lucas proposed updating the color coding for positive and negative error neurons to maintain consistency with previous figures, and Martin suggested ensuring that panel A remains at the algorithmic level, not implying synaptic connections, to avoid confusion.

Suggestions for Figure Structure and Future Work: Stefan proposed separating local circuit structure from inter-area connections using distinct boxes or schematic elements, and the group, including Lucas and Jerome, agreed to consider this approach, with Stefan volunteering to provide a sketch and the team planning to revisit the discussion in the next meeting.

Next Steps and Action Items for Figure Development: Jerome coordinated with Lucas, Stefan, Benedicte, and others to assign follow-up tasks, including preparing sketches and proposals for Figure 2, clarifying algorithmic versus implementation levels, and planning to reconvene for further discussion and consensus.

Assignment of Sketch Preparation: Jerome asked Stefan and Lucas to collaborate on preparing several schematic proposals for Figure 2, with Benedicte available to assist, emphasizing the need for visual sketches to facilitate the next discussion.

Clarification of Figure Levels: Martin recommended clearly distinguishing between algorithmic (abstract) and implementation (circuit) levels in the figures, suggesting the use of arrows or classic circuit symbols in panel A and reserving synaptic details for panel B.

Plan for Next Meeting: The group agreed to revisit the topic in the following week, aiming to finalize the figure based on the new proposals and to ensure the text accompanying the figure accurately reflects the range of possible implementations.

Review Context and Editorial Process: Hannah inquired about the context of the figure revisions, and Jerome clarified that the updates were initiated internally to improve figure harmonization, not in response to reviewer requests, and that the ongoing discussion was triggered by internal uncertainties.

Origin of Figure Revisions: Jerome explained to Hannah that the figure revisions were undertaken by the team, including Colleen, to improve consistency and style across the review, and that the current debate arose from internal discussions rather than external reviewer feedback.