Meeting Notes

• Date: 2025-08-19
• Time: 09:00 (PT)
• Location: Teams Meeting
• Presentations:

Agenda

1. Presentation from Dr Bosman on this recent publication

Meeting Recording

Meeting Notes

Presentation of Predictive Context Effects on Orientation Coding in V1: Conrado Bosman Vittini presented recent research on how predictive context can reshape orientation coding in V1, discussing experimental design, key findings, and methodological details, with active engagement and questions from Jerome, Alexander, Lucas, and Jorge.

Research Motivation and Background: Conrado Bosman Vittini introduced the study by outlining the classical view that orientation preference in V1 is stable, but highlighted evidence that cortical coding is context-dependent, referencing predictive coding models and the oddball paradigm as frameworks for understanding context effects.

Experimental Design and Methods: The team used a modern oddball paradigm with passive presentation of oriented gratings to awake animals, employing laminar electrophysiology with 32-channel NeuroNexus probes to record responses across cortical layers, and included controls such as mini-standards to disentangle mismatch negativity and deviant detection effects.

Key Findings on Layer-Specific Dynamics: Analysis revealed robust stimulus-specific adaptation and mismatch negativity across all layers, with deviant detection significant only in supragranular layers during early phases and in infragranular layers during late phases, as shown by both ERP and spiking activity.

Population Coding and Variance Analysis: Population decoding accuracy increased for deviant conditions at early latencies, and Manhattan distance was used to quantify trial-by-trial variance, showing that deviant stimuli triggered larger shifts in population activity compared to standards, with further analysis linking these shifts to changes in orientation preference.

Orientation Preference Switching: A subset of neurons, particularly in infragranular layers, exhibited context-dependent switching of orientation preference, with these switches occurring predominantly during deviant presentations and late latencies, and further controls indicating that these effects were not due to slow adaptation or block structure.

Methodological Discussion and Clarifications: Alexander and Lucas engaged Conrado Bosman Vittini in detailed questions about the choice of variance metrics, decoding approaches, and the interpretation of trial-by-trial variance, leading to clarifications on the rationale for using Manhattan distance and the specifics of the decoding analysis.

Choice of Variance Metric: Alexander asked why Manhattan distance was chosen over Euclidean distance for quantifying population variance, to which Conrado explained that Manhattan distance is less sensitive to outliers and allows additive contributions per neuron, making it suitable for their analysis.

Trial-by-Trial Variance Justification: Conrado clarified that trial-by-trial variance was used to reliably characterize variability across all cells, acknowledging that other approaches could be explored but that their choice was based on intuition and the results obtained.

Decoding Analysis Details: In response to Alexander's questions, Conrado confirmed that decoding was performed on a trial-by-trial basis, using algorithms to determine whether population activity could distinguish between conditions, and noted that significant decoding accuracy was found for deviant and standard stimuli.

Interpretation of Context-Dependent Orientation Tuning and Field Debate: Alexander, Jorge, and Jerome discussed the implications of the findings for the debate on orientation tuning stability across cortical layers, referencing historical and recent studies, and considering the flexibility of orientation tuning in light of both hardwired and context-driven models.

Historical Context and Layer-Specific Tuning: Alexander referenced past studies showing abrupt changes in orientation tuning at the layer 4-5 boundary and discussed the debate over whether orientation columns are truly stable, noting that increased variance in infragranular layers has been observed in both historical and recent data.

Implications for Hardwired vs. Flexible Models: The group discussed how the observed context-dependent switching of orientation preference challenges the classical hardwired model of orientation tuning, suggesting that cortical circuits exhibit more flexibility than previously thought, with possible contributions from feedback and inhibitory mechanisms.

Suggestions for Further Analysis: Alexander and Jorge suggested that further analysis of orientation tuning curves and selectivity indices across layers could provide additional insights, and Conrado agreed that this would be a valuable direction for future work, referencing supplementary figures and acknowledging the need for more detailed laminar analysis.

Mechanistic and Experimental Considerations: Jerome, Lucas, and Alexander raised questions about the mechanisms underlying context-dependent orientation switching, the potential role of inhibitory neurons, and experimental controls for adaptation and block effects, with Conrado providing explanations and outlining limitations.

Potential Mechanisms for Switching: Jerome asked about possible mechanisms for context-dependent switching, such as involvement of inhibitory cell types or feedback from higher areas, and Conrado responded that both inhibition and feedback are plausible but could not be directly tested in the current dataset.

Control for Adaptation and Block Effects: Lucas inquired about the possibility of adaptation effects bleeding across blocks, and Conrado explained that comparisons between standards and deviants within blocks, as well as inter-stimulus intervals, were used to minimize and control for such effects.

Role of Inhibitory Neurons: Jerome and Conrado discussed the importance of including inhibitory neurons in future analyses to better understand the network mechanisms underlying early versus late phase tuning changes, noting that the current dataset had too few inhibitory neurons for robust analysis.

Future Directions and Ongoing Experiments: Jerome outlined plans for upcoming experiments involving neuropixel recordings and imaging to further investigate context effects, and invited participants to contribute to experimental design and data analysis discussions in future meetings.

Planned Experiments and Stimulus Design: Jerome described ongoing efforts to finalize the stimulus design for experiments with four different context types, including sensory-motor oddball contexts, and noted that pilot experiments with mice on a semi-virtual reality system are underway.

Data Analysis and Collaboration: Alexander mentioned progress and issues with new data files and preprocessing, while Jerome encouraged continued collaboration and engagement on the GitHub forum and in future meetings to refine analysis approaches and experimental plans.