Abstract—In super aged society, population with age related disease such as dementia increases, which is disorder of brain functions. Interactive group conversation or group activities play an important role of activating brain functions. In order to make interactive group conversation among older people for preventing or inhibiting the progress of mild cognitive decline and dementia, our research aims to develop a system assisting group conversation, which is capable of detecting laughter and emotions, and recognizing speech contents of the participants. This study is a part of the developing system to recognize laughter from acoustic signals of group conversation. Mel frequency Cepstral Coefficients (MFCCs) are calculated and vector quantized from the acoustic signals. Minimum distance classifier is applied for recognition of laughing and non-laughing states. Experimental results show very good performance. F-score and accuracy rate are 88% and 98%, respectively.