Although experimental work consolidated this abstraction for basal and proximal apical dendrites, a rigorous test for tuft dendrites is still missing. Recent studies have proposed that individual dendritic branches or subtrees may function as independent computational subunits. However, it is still a matter of debate how pyramidal neurons transform their synaptic inputs into spike outputs. Revealing how these cells operate is key to understanding the dynamics and computations of cortical circuits. Pyramidal neurons are the principal cell type in the cerebral cortex. The iso-response framework proposed in this computational study is highly efficient and could be directly applied to biological neurons. Our findings support the view that dendrites of pyramidal neurons possess non-linear analog processing capabilities that critically depend on the location of synaptic inputs. In contrast to these findings for precisely timed inputs, we show that neuronal computations based on firing rates can be accurately described by purely feedforward two-layer models. Individual dendritic branches bidirectionally modulate the thresholds of their input-output curves without significantly changing the gains. We find that additive feedback alone explains the somatic responses to synaptic inputs to most of the branches in the apical tuft. Then, we relax the core assumption of subunit independence and introduce non-linear feedback from the output layer to the subunit inputs. Using a detailed compartmental model of CA1 pyramidal neurons and a novel theoretical framework based on iso-response methods, we first show that somatic sub-threshold responses to brief synaptic inputs cannot be described by a two-layer feedforward model. It is an open question whether this mathematical abstraction can be applied to apical tuft dendrites as well. The outputs of the subunits are linearly summed and passed through a final non-linearity.

Basal and proximal apical dendrites have been shown to function as independent computational subunits within a two-layer feedforward processing scheme. Dendrites of pyramidal cells exhibit complex morphologies and contain a variety of ionic conductances, which generate non-trivial integrative properties.

Dendrite function