Dendritic integration and apical amplification
Explanation
Most popular theories of consciousness operate at the scale of large brain networks: global workspaces, integration of information, global oscillations. A growing group of neuroscientists argues, in contrast, that the key is at a much more microscopic level: the behaviour of individual neurons, in particular layer-5 pyramidal cells of the neocortex and their apical dendrites. The Dendritic Integration Theory (DIT) of Talis Bachmann and colleagues, and the apical amplification theory associated with Matthew Larkum, articulate this hypothesis.
The physiology is specific. Layer-5 pyramidal neurons have two computational compartments: the cell body (soma) integrates sensory inputs in lower cortical layers, and the apical dendrites —prolongations rising to the superficial cortical layers— receive descending feedback inputs from hierarchically higher regions. When both flows coincide temporally, a dendritic event occurs (plateau potential or BAC firing) that enormously amplifies the neuron's activity. That coupling between soma and apical dendrite would be, according to DIT, the microcircuit mechanism of consciousness.
The intuition is elegant: consciousness would require the coincidence between the ascending signal (what is arriving sensorially) and the descending signal (expectations, context, predictions). When both align in the pyramidal neuron, conscious experience emerges as a state in which sensory content is integrated with cognitive context. When descending feedback is disconnected —by anaesthesia, for example— the mechanism switches off and consciousness disappears, even if ascending sensory activity persists. Various studies in rodents and primates seem to support this idea.
Apical amplification, conceptually close, emphasises the role of apical dendrites as a contextual switch: they control whether a sensory input is processed consciously or not, depending on whether there is concurrent apical amplification. This proposal connects naturally with predictive processing: descending expectations modulate ascending processing precisely at that dendritic interface. Larkum and collaborators have worked intensively on electrophysiologically characterising these mechanisms.
A recent 2025 review treats DIT and apical amplification explicitly as circuit-level theories of consciousness, distinguishing them from large-scale theories such as IIT or GNWT. This reflects an interesting shift in the field: while dominant theories operated at a high description level (networks, information, access), there is an emerging current trying to anchor consciousness in specific cellular mechanisms, susceptible to pharmacological, genetic, optogenetic intervention. It is an attempt to make the theory of consciousness mechanistically tractable at the level where one experiments with rodents in the laboratory.
The main weakness today is exactly that: they promise much at the mechanistic level, but still have less direct human evidence and less comparative maturity than large-scale theories. Most empirical work is in rodents and non-human primates; transferring the findings to subjective human consciousness requires delicate inferences. Furthermore, apical dendrites are possibly a necessary mechanism —but sufficient? An isolated pyramidal neuron with BAC firing does not seem to produce experience; the step from microcircuit mechanism to global phenomenality still requires work. Even so, DIT and apical amplification represent a promising line that any updated catalogue should cover.
Strengths
- Concrete and testable microcircuit mechanism.
- Explains why anaesthesia disconnects consciousness while maintaining sensory activity.
- Convergence with predictive processing at the apical interface.
- Opens pharmacological and optogenetic intervention as empirical routes.
- Complementary with large-scale theories.
Main critiques
- Direct evidence in humans still limited.
- Going from microcircuit mechanism to global phenomenality unresolved.
- Risk of describing a necessary condition without sufficiency.
- Difficult integration with non-local theories (IIT, GNWT).