This work trains a Deep Neural Network (DNN) based model of flicker fusion with human psychophysics data. The convolution filters of DNNs trained on natural images acquire the features of gabor filters. Similarly the convolution filters of the DNN trained with temporal psychophysics data acquired symmetrical features. Derivatives of gaussians and gabor functions found in human visual systems are often symmetric. The predictions made by the DNN on a complex flicker stimulus was tested with psychophysics experiment. The model is shown to be falsifiable and can be improved with further training.