This thesis investigates psychophysiological characteristics of false memories in two groups of people: healthy participants and amnesic brain injured patients with Korsakoff syndrome (KS). KS patients present impairment of diencephalic structures in the brain, mainly at mammillary bodies, thalamic nuclei, fornix and mammillothalamic tract as consequence of thiamine deficit. This deficit is associated with dense anterograde amnesia with general preservation of other cognitive abilities. The main objective was to study how patients processed false memory compared to healthy agematched controls. To do it so, we produced several experiments. A first one was designed to validate a new false memory task that offers a language-free alternative to the classic Deese- Roediger-McDermot paradigm (DRM) along with some other improvements introduced to bias false memory production. We tested this visual false memory task (VFMT) in a sample of 20 healthy participants to study its validity as true and false memory generator under event-related potential (ERP) conditions. Results indicated that classic ERP old/new effect was present despite in showed a consistent central localization in the scalp compared to what previous ERP experiments found. Moreover, we found that true and false memory ERP signal appeared to be equivalent in localization but different in voltage. We aimed to study true and false memory in amnesic patients and compare their performance with age-matched healthy controls. But before doing it we run an experiment to adapt VFMT to amnesic characteristics of KS sample. We decreased the amount of information to be studied engaging a shorter-delay testing time that may overcome episodic and attentional difficulties in patients. Second version of the task, i.e. VFMT2.0 was behaviourally applied to a sample of 10 KS patients to confirm that this new version successfully produced enough amounts of true and false memories. Neuropsychological assessment was also applied to KS patients to quantify and characterize their cognitive impairment and brain damage. Finally, ERP experiment was performed in KS patients and age-matched controls. Respect to results from first version of VFMT, ERP differences appeared in healthy participants. What behaviourally was reflected as an easier memory task, it also was associated with ERP differences in localization of ERP activity in the scalp, implicating more frontal-located electrodes for true memory processing compared to the central distribution of ERP found at VFMT1.0. Moreover, under VFMT2.0 task, healthy participants showed ERP differences when true and false memories were compared: right-frontal electrodes for true and left-frontal for false memory processing on 500 to 1000ms time window. Regarding amnesic patients, brain activity associated to true and false memory was equivalent and left-frontal sited all over the epoch. Main results indicated that true memory processing showed different ERP characteristics when patients were compared with controls despite their behavioural false memory rates were equivalent. When false memory was analysed, differences between groups were also found, mainly at early 300-500ms and later 1000 to 1500ms time windows.