Recursive mentalizing and public salience in the perception of common knowledge Common knowledge—what everyone knows that everyone knows, ad infinitum—is theoretically required for coordination. But a finite mind cannot represent infinitely embedded propositions, and even a few are cognitively taxing. We propose that people perceive common knowledge by combining a capacity-limited ability for recursive mentalizing with an intuitive sense of public salience: When an event is jointly and openly witnessed, they infer that recursively embedded beliefs are licensed. We probed participants’ attributions of embedded beliefs to characters in fictitious vignettes. Study 1 presented narratives in which the information available to characters was either asymmetrical, reciprocal, or public. Participants indicated their confidence in the characters’ beliefs with 0, 2, 4, 6, and 18 embeddings (the latter judged indirectly). Ratings showed i) a decline with number of embeddings, reflecting cognitive load, ii) modest overattribution of embedded beliefs in the nonpublic conditions, iii) far higher attribution of embedded beliefs in the public conditions, and iv) high indirect judgments that arbitrarily deep beliefs were justified. Study 2 replicated these findings with animated videos in which the characters’ information was private, reciprocal, doubly reciprocal, or public. Together, the studies indicate that people attribute common knowledge to observers of a publicly salient event, while being susceptible to capacity-driven recursion collapse and heuristic extrapolation (false embedded beliefs). This can explain how humans achieve coordination from sparse cues, yet sometimes miscoordinate when publicness is absent or embedding grows complex.
Strong mnemonic prediction errors increase cognitive control, attention, and arousal Ongoing experience is continuously processed in the context of past events. Divergence between current experience and memory-based predictions (i.e., a mnemonic prediction error; MPE) is theorized to be a signal for the hippocampus that encoding, as opposed to retrieval, should be prioritized. We asked how MPEs place demands on cognitive and neural resources beyond the hippocampus, and whether these demands differ as a function of prediction strength. We investigated these questions across two experiments, wherein we recorded scalp electroencephalography and/or pupillometry as 101 young human adults performed an associative memory task. Strong MPEs, more so than weak MPEs, increased physiological indices of cognitive control (frontal theta), attention (posterior alpha and pupil size), and arousal (pupil size). Trial-level pupil-linked MPE responses scaled with the amount of attention (posterior alpha) allocated during prediction generation. Finally, greater cognitive control (frontal theta) during strong MPEs promoted better learning of prediction violating (i.e., unexpected) stimuli. Collectively, these findings reveal multiple pathways through which the mind and brain respond adaptively to violations of strong mnemonic predictions.
Integrated information theory (IIT) and the testability of the silent neuron predictions Integrated information theory (IIT) makes two predictions about the role of inactive neurons in consciousness. According to the silent brain (SB) prediction, rendering all active neurons inactive (“silent”) in the physical substrate of consciousness (the “main complex”) does not eliminate the presence of consciousness, because the neurons are still able to spike. According to the disabled neuron (DN) prediction, rendering a subset of silent neurons in the main complex no longer able to spike (“disabled”) can impact the qualitative character of experiences “nonconventionally” associated with those neurons. Bartlett (2022) argues that these predictions are untestable, because evidence for either prediction would imply that the testing conditions were not met. In this paper, we provide a detailed analysis of both silent neuron predictions, showing how they can in fact be tested. For the SB case, we clarify how a neural mechanism outside of the main complex can yield the required report of consciousness while maintaining the SB state. For the DN case, we distinguish between two ways of explaining how a neural mechanism could casually interact with the main complex: an IIT-inspired “dispositionalist” explanation, and a more conventional “actualist” explanation. Drawing on the work of Imre Lakatos, we conclude with a discussion of how the distinction between the two explanations sheds light on why it is so difficult to resolve theoretical disputes about consciousness. Despite these difficulties, we provide a framework that can lead to concrete progress for consciousness science.
The grounding of abstract concepts in the sensory-motor, mentalizing and social interaction brain systems: A multiple representation framework For decades, theorizing in the cognitive sciences was dominated by the assumption that abstract concepts, lacking directly perceivable referents, can only be represented and processed through amodal or verbal linguistic representations. More recently, refined hybrid grounded cognition theories have successfully been applied to account for abstract concepts within a multiple representation framework. According to this perspective, the meaning of abstract concepts is constituted by representations distributed across various experiential modal brain systems, including sensory, motor, emotional, mentalizing, and social interaction networks. Representations in these modal systems are complemented by representations in language-related and amodal hub regions, depending on the specific semantic content of the concept. In this article, we first outline a multiple representation framework within a hybrid grounded cognition approach and then review neuroscientific evidence concerning the neural substrate of abstract concepts related to sensory-motor features, mental states and social constellations. These findings indicate that modal sensory-motor, mentalizing, and social interaction brain systems contribute to the processing of particular types of abstract concepts, alongside representations in amodal semantic hub regions and language areas. Moreover, this body of research demonstrates that different modal neural circuits are engaged as a function of a concept’s specific semantic content, thereby highlighting the heterogeneity of abstract concepts. Future research should address outstanding questions, including the precise functional contributions of distinct neural circuits to the representation of abstract concepts and the evaluation of predicted patterns of impairment in neurological and psychiatric patient populations.
Cross-region neuron co-firing mediated by ripple oscillations supports distributed working memory representations High-frequency (~90-Hz) ripple oscillations may promote integrative processing in mammalian brains. Co-occurrence of ripple oscillations has been associated with enhanced temporal binding of neural activity between nearby human cortical neurons, but whether co-ripple facilitation of neuronal coupling supports cognitive processing or occurs at greater distances remains unclear. Here we analyze intracranial recordings from patients implanted with microwire electrodes in the hippocampus, amygdala, ventromedial prefrontal cortex, anterior cingulate cortex and pre-supplementary motor area, bilaterally, during a working memory task. We demonstrate that ripple rates increase in all recorded regions during encoding, maintenance and retrieval. Co-occurrence of ripples increases between brain regions, associated with ~30% increases in cross-region co-firing, without decrement over distances up to 220 mm. Cross-regional co-rippling and associated co-firing scale with memory load during maintenance and retrieval. During retrieval, co-ripples promote reinstatement of stimulus-specific, long-distance co-firing patterns observed during encoding, especially during rapid recognition. Co-occurring ripple oscillations thus coordinate long-range, stimulus-specific neural co-firing supporting distributed representations during human cognition.
Extensive Experience Remodels Neural Task Circuitry to Escape the Frontal Bottleneck and Increase Automaticity of Categorization Object category learning is a foundational cognitive process. Most human category learning studies involve brief paradigms lasting a few hours and show increased shape tuning in visual areas and task-dependent responses in PFC. Other studies also identify a “frontal bottleneck” that limits multitasking. However, real-world categorization often involves months or years of practice, potentially producing qualitative shifts toward automaticity. We tested the hypothesis that extensive training causes a spatio-temporal shift in the neural circuitry supporting categorization. Participants were trained over >30,000 trials across 5–10 weeks to categorize novel morphed car stimuli via a mobile app. We used fMRI and EEG rapid adaptation techniques to examine neural responses after initial learning (∼4 hr in 1–2 weeks) and after extensive training (∼16 additional hours over another 4–8 weeks). Converging fMRI and EEG results showed that extensive training fundamentally remodeled task-related circuitry: Visual areas in ventral occipito-temporal cortex (vOTC) were initially shape-selective, but category-selective responses emerged in the vOTC after extensive training. The vOTC also showed decreased functional connectivity with the PFC and increased connectivity with motor output areas. This supports the hypothesis that extensive experience enables category decisions to occur outside of the “frontal bottleneck.” Critically, the decrease in connectivity between vOTC and PFC was associated with improved categorization performance while dual tasking, indicating increased automaticity. These findings demonstrate that prolonged training reshapes the neural basis of categorization, shifting it from a flexible but attentionally controlled process to a more streamlined, automatic process.
Distinct High-Gamma Signals in Primate Prefrontal Cortex Differentiate Cue Information, Preference Revision, and Expected Reward During Value-Based Decisions Understanding how prefrontal cortex (PFC) signals evolve over time to support decision-making requires characterizing both the spectral and temporal structure of neural activity. High-gamma (Hγ) power in local field potentials (LFPs) reflects local population firing, yet its role in differentiating cue-driven preference revision, reward expectation, and outcome-related signals during reward-guided decisions remains unclear. We recorded LFPs from the anterior cingulate cortex (ACC), dorsolateral PFC (DLPFC), and orbitofrontal cortex (OFC) in macaques performing a multicue reward-based decision task in which up to four sequential cues indicated the expected reward associated with competing targets. Hγ power was extracted in sliding windows aligned to key task events, and support vector machine classifiers were used to decode cue-driven preference revision, reward expectation, cue value level and position, target choice, and prediction-error contrasts. Hγ responses robustly differentiated preference-reversal from confirmation trials, with subject-specific modulation patterns across PFC subregions. Cue value level and spatial position were reliably decoded across PFC, with stronger differentiation of cue value level in OFC and cue position in DLPFC. Around movement onset, Hγ activity differentiated high versus low expected reward across PFC, with decoding accuracies reaching 85% in ACC. Hγ was also modulated by unexpected reward omission, yielding 78% decoding accuracy from a single OFC channel. These findings demonstrate that Hγ activity across PFC subregions differentiates multiple computationally defined decision variables with distinct temporal profiles and overlapping regional contributions, highlighting Hγ as a robust marker of dynamic reward-guided decision processes.
Dynamics of Dentate Gyrus Place Cells and Dentate Spikes during Spatial and Nonspatial Changes in Environments The dentate gyrus (DG) is thought to play a key role in the formation of dissociable memory representations for similar contexts. Neurons in the DG receive highly processed spatial and nonspatial sensory information from the medial and lateral entorhinal cortices, respectively. Changes in spatially tuned firing patterns of DG place cells occur after spatial changes to an environment, but the degree to which DG place cells respond to ethologically relevant nonspatial stimuli is largely unknown. Spatial and nonspatial information is thought to be transmitted to the DG during discrete local field potential events called dentate spikes. Here, we tested the extent to which different spatial and nonspatial stimuli modulate place cell firing patterns and dentate spike dynamics. We performed extracellular recordings of DG place cells and local field potentials in rats of both sexes exploring a familiar spatial environment, in which social stimuli and nonsocial odors of varying ethological relevance were presented, and a novel spatial environment. As expected, DG place cells exhibited different firing patterns between familiar and novel environments. Significant changes in firing were not observed, however, with any of the nonspatial stimuli. Surprisingly, the occurrence of dentate spikes associated with lateral entorhinal cortex input increased during exploration of ethologically relevant stimuli, and this increase was greater for social stimuli. Altogether, these results suggest that the DG preferentially responds to social stimuli at the network level, providing novel insights into how spatial and nonspatial information is processed in the DG.