Neural computation depends on inter-areal signal transformations, which are determined by the input-output (I/O) functions of individual and networks of neurons. Due to intrinsic neuronal properties and inter-neuronal interactions, networks can show preferences for synaptic inputs in certain frequencies, e.g. in the form of lowpass filtering or resonance [Izhikevich 2003, Cardin et al 2009, Lewis et al 2021, Pike et al 2000]. High-density silicon-probe recordings were made from area V1 and V2 in awake mice. We used optogenetic stimulation (continuous (1s), pulses (5ms), sinusoids) with an opsin having fast kinetics (Chronos) to gain precise temporal control over specific cell populations and to study the I/O functions of (1) V1 neurons expressing the opsin, (2) excitatory and inhibitory V1 and V2 neurons not expressing the opsin. We determined the dependence of firing rates, phase-locking, coherence and power on stimulation frequency. Spike-laser phase-locking increased steeply towards higher frequencies, indicating that opto-tagged excitatory neurons high-pass filtered optogenetic inputs. This was explained by the narrowing of phase distributions with frequency, suggesting a non-linear I/O transformation. Fast-spiking interneurons closely followed the phase-locking of excitatory neurons without exhibiting resonance. Surprisingly, non-opto-tagged excitatory neurons, both in V1 and V2, phase-locked predominantly to low-frequency optogenetic inputs. Next, we compared different measures of I/O transformations. Strikingly, opto-tagged excitatory neurons responded with similar firing rates to all optogenetic input frequencies. Likewise, spike-field coherence was relatively flat, indicating optogenetic inputs were reliably encoded at all frequencies. Thus, in the absence of synaptic filtering, neurons encode and respond to different frequencies very similarly, despite being more phase-locked to high frequencies. Together, these findings indicate a major difference between the filtering of synaptic and optogenetically-induced inputs. They further suggest that area V1 does not exhibit band-pass resonance and that local as well as inter-areal synaptic communication is most effective at low frequencies.