This paper aims to establish a method for detecting major depressive episodes experienced by people with bipolar or major depressive disorder. The method only considers motor activity and total sleep time per day. Both symptoms were measured using actigraphy. Motor activity and total sleep time are usually evaluated subjectively according to the patient’s statements or a professional’s perception. However, it is better to use a linguistic description of the patient’s situation based on objective measurements. As a test bed, we used a public domain database concerning actigraphy data of merely 55 participants that either have or do not have one of the two aforementioned mood disorders. We resorted to fuzzy natural logic (FNL) since this theory takes linguistic expressions and human reasoning into account and seeks logical conclusions that are applicable to complicated and real situations, thus being a resource that helps to develop diagnostic reasoning and not replace it. Initially, we created implicative rules based on already established knowledge about mood disorders and designed fuzzy sets that describe the total sleep time and motor activity of each participant in terms of evaluative linguistic expressions (ELE). Using intermediate quantifiers, we were able to reduce the number of linguistic rules from fifteen to three.