Neural activity in the cortex exhibits a wide range of timescales, allowing animals to respond to rapidly changing sensory stimuli while maintaining information in working memory. The timescales of neural activity are not only affected by bottom-up sensory inputs, but also by top-down inputs originating from corticocortical and neuromodulatory pathways. In addition, cortical circuits comprise a large diversity of cell types connected in non-random ways, which are the subject of profound plasticity dynamics. Due to this complexity, understanding how top-down projections affect the timescales of cortical network dynamics remains elusive. Most existing models assume uniform connectivity architectures that allow analytical treatment but are inconsistent with biologically measured connectivity emerging from well-coordinated synaptic plasticity mechanisms. Here, we investigated how networks with connectivity generated by experimentally measured synaptic plasticity rules respond to different forms of top-down modulation. We trained large-scale spiking networks with multiple inhibitory interneurons (parvalbumin PV, somatostatin SST and vasoactive intestinal peptide VIP) using different spike-timing-dependent plasticity (STDP) rules at the different synapses. Depending on the combination of plasticity rules, we found that network connectivity stabilizes to two scenarios. Specifically, by incorporating or omitting STDP rules at SST-to-excitatory neuron synapses, the circuits can be primarily dominated either by disinhibition (Fig. 1A) or inhibition (Fig. 1B) of excitatory cells, resulting in different dynamical regimes. In the two networks, different forms of top-down modulation can slow down (Fig. 1C) or speed up neural activity (Fig. 1D), affect the speed of stimulus encoding (Fig. 1E) and shift the networks between being primarily recurrently driven or feedforward driven (Fig. 1F), thereby providing the opportunity to promptly adapt to external or internal representations. We show how these two networks can coexist within a single biologically plausible cortical circuit while retaining all the individual properties of each network. This work highlights that different plasticity mechanisms structure networks to process information at a range of timescales and adapt to a rapidly-changing environment.