The response of sensory neurons to different stimuli are largely characterized by tuning curves, with a peak of activity at a preferred stimulus. Tuning curves are thought to be shaped by Hebbian plasticity during development through the strengthening of synapses between neurons with similar functional preferences. This raises the hypothesis that the strength of synapses should correlate with the difference in functional preference between pre- and postsynaptic activity. A recent study, however, found that the functional preference of cortical neurons was correlated with the number of spines activated by a given stimulus, rather than with their strength [1]. Furthermore, they found that synaptic strength actually correlated with spine selectivity. Here, we study what type of plasticity rule could lead to the emergence of such connectivity pattern and what are the functional implications in a sensory circuit. We use a rate model to simulate a single neuron receiving input from multiple presynaptic sources with different preferred stimuli and different selectivity. We find that a plasticity rule based on presynaptic activity only, more specifically the variance of presynaptic activity, leads to the formation of a feed-forward circuit where the strength of individual synapses is correlated with presynaptic selectivity, but independent of the presynaptic preferred stimulus. Assuming selectivity to be a measure of input signal reliability, we further explore how the correlation between synaptic weight and presynaptic selectivity could be relevant in the context of Bayesian cue integration. Our results suggest a model of sensory cortex where the number of spines defines preferred stimulus, while the strength of individual synapses can be used as a measure of input reliability.