A phototube is connected to a simple neural circuit for developing a light-sensitive neural circuit, which can be regulated to trigger suitable firing modes by taming the frequency or amplitude of external stimulus. Furthermore, the photoelectric neuron and its Hamilton energy are obtained through scale transformation. It is discerned that the firing modes of neuron is dependent on the energy level of the neuron. Chaotic signals are filtered and encoded to excite the neuron, then chaotic firing modes are obtained and the modes transition and energy value are changed significantly by increasing the intensity of the filtered signals from chaotic system. Considering the complexity and plasticity of the synapse, a nonlinear resistor is used to couple two photosensitive neural circuits. The synchronization stability and energy exchange between two coupled neurons presenting with different initial firing modes are investigated by applying periodic and filtered signals on the neurons. Furthermore, additive noise is applied to investigate the synchronization stability and energy propagation between the two neurons. When photocurrent is selected in periodic type, two coupled neurons are blocked to reach complete synchronization, the coupling intensity is fluctuated with time and energy balance is broken between two neurons. By taming the noise intensity, two bursting neurons can reach complete synchronization while two chaotic neurons seldom reach synchronization when the neurons are driven by peridic currents. When filtered signals are used to excite the two coupled neurons, intermittent phase lock and phase synchronization can be realized between two neurons.