TL;DR
This study reveals that when marmosets actively engage in a sound localization task, their auditory cortex neurons respond not only to sounds within their usual receptive fields (RFs) but also to sounds originating from outside these RFs. This "off-RF" facilitation, particularly prominent in caudal auditory areas, suggests that behavioral demands dynamically reshape neural representations, extending beyond static RF boundaries.
What Problem Does This Paper Address?
The established view in sensory neuroscience is that neurons have specific receptive fields (RFs) – areas of sensory space that, when stimulated, elicit a response. While it's known that attention can enhance responses to stimuli within these RFs, how attention influences responses to stimuli outside the RFs, especially in the auditory system, has been less clear. This gap is crucial because real-world auditory tasks often involve processing sounds from various locations, not just those precisely matching a neuron's predefined RF. The authors aim to understand if and how behavioral engagement dynamically alters the spatial tuning of auditory cortex neurons, particularly exploring responses to sounds originating from outside their classical RFs.
Methods
The researchers trained two marmosets to perform a spatial discrimination task. They recorded single-unit activity in the auditory cortex while the animals listened passively to sounds from a full spherical array of speakers and while they actively performed the discrimination task. Key techniques included:
- Full-field spatial RF mapping: Using a dense array of 24 speakers, they mapped the spatial RFs of neurons during a passive listening condition.
- Behavioral Task: A Go/No-Go task where animals had to lick for a reward when a target sound was presented and withhold licking for background sounds.
- Stimulus Paradigm: A novel "random spatial profile" stimulus was used for intracellular recordings to reveal suppressed responses below the spiking threshold.
- Computational Modeling: A two-dimensional normalization model was employed to simulate the observed neural responses, incorporating stimulus, attention, and suppression fields.
- Intracellular Recordings: Used to measure membrane potentials and reveal sub-threshold suppressive effects.
Datasets
The study utilized data collected from two marmoset monkeys. No external datasets were used for training or evaluation. The data collected from these animals, encompassing neural recordings across various auditory cortex regions during passive listening and active task performance, forms the basis of the study.
Key Insights and Ideas
- Dynamic RF Expansion: Behavioral engagement doesn't just amplify responses within existing RFs; it actively recruits neurons to respond to sounds originating from outside their classical RFs. This suggests that RFs are not static but are dynamically reconfigured by task demands.
- Caudal Dominance: This "off-RF" facilitation was significantly more prevalent in the caudal auditory cortex (CM/CL) compared to the rostral auditory cortex (R/RT) and the primary auditory cortex (A1). This points to regional specialization in how auditory space is represented and modulated by attention.
- Widespread Suppression as a Mechanism: The authors propose and model that widespread suppression, rather than just excitation, plays a critical role in this phenomenon. This suppression, particularly outside the RF, might allow for a more focused and enhanced response to the attended stimulus.
- Task Performance Matters: The observed neural modulation was strongly correlated with successful task performance (hits), suggesting it's tied to actual auditory processing for the task rather than general arousal or motor preparation.
Key Results
- Off-RF Facilitation: A significant proportion of neurons showed increased firing rates to target sounds presented outside their passive RFs. In some cases, these off-RF responses even exceeded responses at the RF center during passive listening.
- Regional Differences: Task-related modulation (measured by the modulation index, MI) was significantly larger in CM/CL compared to A1 and R/RT.
- Model Validation: The normalization model successfully reproduced the experimental findings, predicting that widespread suppression is a necessary component for explaining the observed off-RF facilitation.
- Sub-threshold Effects: Intracellular recordings revealed hyperpolarization (suppression) of membrane potentials at off-RF locations, even in neurons that didn't show suprathreshold firing changes, highlighting the role of inhibitory processes.
- Performance Correlation: Neurons with enhanced responses (positive MI) showed higher behavioral sensitivity (d') compared to those with suppressed responses (negative MI), indicating a link between neural modulation and task accuracy.
Limitations and Future Work
- Species Specificity: While marmosets are a good model, direct comparisons to other species, particularly humans, would be beneficial.
- Stimulus Order Control: While a control condition was used to mitigate stimulus order effects, the authors acknowledge that it's challenging to completely disentangle all potential confounds.
- Model Simplification: The normalization model, while powerful, is a simplification of complex neural circuits. Future work could explore more detailed circuit models.
- Further Behavioral Tasks: Investigating these effects across a wider range of auditory tasks and behavioral demands would provide a more comprehensive understanding.
- Mechanistic Detail: While suppression is implicated, further research is needed to fully elucidate the specific neural circuits and neurotransmitter systems involved in generating this widespread modulation.
Summary updated