A long-standing question in cricket neurobiology is what kind of neuronal mechanism the females use, to recognise the temporal pattern of the male calling song. A delay-line and coincidence detector circuit with five neurons underlying auditory pattern recognition in the bi-spotted cricket Gryllus bimaculatus was identified by Schöneich et al (2015). My results here confirmed the morphology of the five identified neurons. Moreover, the group labeling of the auditory neuropil in cricket brain shown in my thesis demonstrated that there are at least 51 neurons in the auditory neuropil. The cell bodies are grouped into six clusters with five in the hemisphere ipsilateral and one in the hemisphere contralateral to the ring structure. In the previous researches, the mechanisms underlying the temporal processing were mostly explored by testing sound patterns with varied pulse period. Recent behavioural tests were performed by systematically varying single pulse durations or inter-pulse intervals, which also led to the design of a chirp pattern with 5-20-50 ms pulses that is attractive when played forward, but non-attractive when the same pulses are played backward as a 50-20-5 ms chirp pattern. In this thesis, I used these sound paradigms to explore the mechanism underlying the processing of individual pulses and intervals by recording the identified neurons in the pattern recognition circuit in cricket brain. The recordings of the pattern recognition circuit reveal the neurophysiological response properties of the five neurons in response to these sound paradigms. The pattern recognition circuit filters the temporal patterns of the chirps progressively. The tuning of the coincidence detector (LN3) and the feature detector (LN4) to the stimuli with varied intervals matched the behavioural tuning but did not reflect the behavioural responses when the duration of pulses were varied. The results also demonstrate that the mechanisms underlying the filtering of pulses or intervals in the pattern recognition circuit may be different. The processing of a train of pulses was accumulative and the activity elicited by a sound pulse in the pattern recognition circuit influences the subsequent responses. One single long interval is sufficient for a reduced activity in LN4 compared to the normal chirp. However, when intervals were kept constant, it required at least three consecutive long pulses for LN4 to respond differently compared to the normal pulses. The attractive and non-attractive patterns elicited a very similar level of activity in the feature detector LN4 in terms of AP/chirp. Further exploration is required to clarify the possible reason for this discrepancy between the neural and behavioural activity.