Vagus nerve stimulation (VNS) is a promising bioelectronic therapy to treat many drug-resistant pathologies. The activity of vagal unmyelinated fibers is an interesting feedback signal to control VNS, since linked to many pathological conditions of the immune, metabolic, cardiovascular, and respiratory systems. However, a widespread belief prescribes that unmyelinated activity is not observable in recordings from implanted electrodes. We tested this hypothesis using a biophysical model of recording setups from the vagus nerve. We considered the cases of a commercial cuff electrode and the transverse intrafascicular multichannel electrode (TIME). The electrogenic activity of myelinated and unmyelinated fibers is solved in NEURON. It is converted into recording signals using the sensitivity matrix of the electrode computed in COMSOL. Simulations revealed that spikes of unmyelinated fibers translate into recording footprints with much more temporal localization when using the TIME electrode compared to the cuff electrode. Consequently, in cuff electrodes the activity of unmyelinated fibers is reflected in the low-frequency band (<<100 Hz) of the recorded signal. Conversely, in TIME electrodes it spans a higher frequency band (100-500 Hz), which is known as local field potential (LFP). This evidence is not in contrast with the experimental literature, which typically relies on extraneural electrodes and focuses on the multi-unit activity band (> hundreds of Hz) of the recorded signal. These results bring attention to the LFP of intraneural recordings from the vagus nerve as a possible proxy of many visceral functions. Experimental validation may pave the way for effective closed-loop protocols of VNS.