Successful integration of feedforward and feedback signals play a major role in cortical processing [1]. Pyramidal cells are thought to be the main sites to perform this integration locally. Signals from the two different pathways are integrated at distinct sites of the pyramidal cell [2]. Information originating up-stream, closer to primary sensory input, is integrated at basal peri-somatic dendrites and drives the cell's firing. Contextual information from down-stream sources is integrated at distal apical dendrites and amplifies the cell's firing only if coinciding with basal input. Importantly, which basal and apical input patterns drive each cell is subject to learning. The brain's learning algorithm needs to rely on local and specific, and global but unspecific components to solve credit assignment and overcome the weight transport problem [3]. How pyramidal cells accommodate for functional processing and learning at the same time is an open question. Here, we propose a rate-based model that combines the processing and learning aspects of pyramidal cell function within a single neuro-dynamical system. Distinctly, it uses the same set of feedforward and feedback signals to implement both, contextual feedforward-feedback integration for processing and goal-directed learning based on structural-temporal credit assignment. Basal-apical integration of feedforward and feedback signals follows an asymmetric input combination pattern, where feedback signals gain-modulate neural activity. The amplification of output activity results in higher-than-baseline firing rates akin to action potential bursts [4]. Besides the impact on immediate activity, this basal-apical coincidence triggers an eligibility tag formation signaling the cell's contribution to current higher-level computation. This tag enters into a four-factor learning rule and takes the role of a structural crediting component [5]. The remaining three factors consist of a basal neo-Hebbian synapse-specific adaptation signal with pre- and postsynaptic components and a global neuromodulatory reward signal. A local network of (dis-)inhibitory interneurons complements the neuron's internal basal and apical integration to help steer the regimes of basal-apical integration. We compare the model's output characteristics to data from a high-detail multi-compartment model of pyramidal cell function [6], discuss its biological relevance, and show that it can learn to improve on a visual decision-making task.