The aim of this thesis work is to investigate whether any existing authentication techniques can be considered as suitable and usable solutions for authentication systems on mobile cognitive devices adapted especially for people with mild dementia. In particular, the study attempts to find out if current authentication approaches may enable protection of access to personal identifiable data on mobile cognitive devices from an unauthorized individual in a manner that does not require intrusive authentication procedures from authorized users- people with mild dementia. The MemoryLane Mobile Cognitive Assistant is taken as an example of a mobile cognitive device designed to help demented elderly who are struggling with memory impairment during their daily life and activities. Due to the nature of the particular research a qualitative research method was selected. In order to collect necessary data an online survey and face to face interviews were conducted. Matrixes and graphs were used as the data analysis methods to draw conclusions and verify them. As a result of the research, based on a literature review and empirical data analysis I found out that the user authentication approach in the form of biometrics, in particular, speaker verification can be considered the most suitable authentication solution from a usage perspective and can be recommended for future implementation on memory supportive ICT devices where the end users are people with mild dementia.