This article takes up the neuroscientific assumption of our brains as "solitary" and contrasts this understanding with the description of actual clinical practices. Drawing on observations of clinical consultations and team meetings in a world famous US center for the diagnosis of dementia, I examine how the "informant", a member of the patient's family, participates in the diagnosis process. Based on specific situations in which the informant is judged to be a "bad" one, I inquire as to how clinicians use what they understand of the affective relationships between the patient and the bad informant in order to make a diagnosis. The diagnosis of dementia in an individual is shown to draw on relational dimensions in the patient's life, made visible and enunciable only when problematic. This inquiry therefore brings out how these neurologists, even though they are engaged in a neuroscientific paradigm that conceives the brain as a self-sufficient cognitive machinery, nevertheless do consider what links us to the brains sharing our lives, in order to make sense of our networks of neurons.