The role of hippocampal atrophy in depression: a neurocomputational approach

Abstract

The role of hippocampal atrophy in the pathogenesis of major depression remains under investigation. Here, we show, within a neural network model, that the incorporation of atrophy reproduces the changes observed in cognitive impairment in depression and could also contribute to the maintenance of the depressed mood. Some other clinical observations, such as treatment resistance and frequent relapses of illness, could also be explained within the framework of the model. We also simulate the action of cognitive therapy and a combined treatment of cognitive therapy and antidepressant drugs. Our findings suggest that, in the presence of hippocampal atrophy, the incorporation of antidepressant drugs would be necessary for the reversal of symptomatology.

Fig. 1

Architecture of the neural network model. a Modules of the network with their population of neurons. An input module (32 neurons) representing perceptual characteristics of a stimulus sends information to a semantic module (32 neurons) and to an affective module (16 neurons), which represent the semantic and the affective content of the stimulus, respectively. There are feedback connections between the semantic and the affective modules. Semantic and affective nodes send information to output semantic (32 neurons) and output affective (16 neurons) nodes, respectively. The names of the weight matrices are shown between brackets. b Connectivity between the nodes of the network. For visual help, only the connections of some representative neurons in each module are shown

Fig. 2

Reaction times for positive, negative, neutral, and overtrained stimuli predicted by the normal and depressed network for the lexical (a) and valence tasks (b). The model reproduces temporal delays of reaction times consistent with human behavior in experiments reported by Siegle [11]. Points in the graphs represent averages over 1,000 random realizations. We follow the same graphical presentation used by Siegle to facilitate comparison. In particular, it is useful to remark that lines in the graphs are only for visual help

Fig. 3

Simulation of HA in the model. Atrophy was implemented on the normal (a, b) and the depressed network (c, d). Four levels of atrophy (5%, 10%, 15%, and 20%) were simulated. For each curve, a new independent set of 1,000 realizations was randomly chosen

Fig. 4

Simulation of cognitive therapy on the depressed network. The relearning curves represent the progress of the therapy. a Lexical task. The arrow indicates the direction of change in reaction times as relearning progresses. b Valence task. The arrow indicates the direction of change in reaction times for positive and neutral stimuli as relearning progresses. Reaction times for negative stimuli move in the opposite direction with the relearning

Fig. 5

Simulation of cognitive therapy on the depressed network with atrophy. The relearning curves represent the progress of the therapy. Atrophy was applied at a level of 10%. a Lexical task. The arrow indicates the direction of change in reaction times for neutral and nonovertrained negative stimuli as relearning progresses. Reaction times for the overtrained negative stimulus move in the opposite direction with relearning. As relearning progresses, reaction times approach values similar to those obtained when a 10% level of atrophy is applied to the normal network. b Valence task. The arrow indicates the direction of change in reaction times as relearning progresses, for positive and neutral stimuli. Reaction times for negative stimuli move in the opposite direction with the relearning. As relearning progresses, reaction times approach normal levels

Fig. 6

Simulation of a combined treatment of psychotherapy and antidepressant drugs on the depressed network with atrophy. The relearning curves represent the progress of the therapy. Atrophy was applied at a level of 10%. a Lexical task. The arrow indicates the direction of change in reaction times as relearning progresses. b Valence task. The arrow indicates the direction of change in reaction times for positive and neutral stimuli as relearning progresses. Reaction times for negative stimuli move in the opposite direction with the relearning. In both tasks, as relearning progresses, reaction times approach normal levels

Fig. 7

Computation of reaction times for the valence task when a positive stimulus enters the network. During the valence task, the network could identify a stimulus as positive, negative, or neutral. For each of these possible responses, counters are defined (C₊: counter for positive valence; C₋: negative valence; C_n: neutral valence). At the beginning of the trial, all counters are set to zero. In every iteration, each counter changes its value according to how well the output affective activation fits the valence pattern that the counter represents, evaluated as the cosine between the vectors (see Eq. 6 in the Appendix). When one of the counters reaches a threshold, the trial ends, and the number of iterations elapsed until that moment is computed as the reaction time of the network. In the figure, a positive stimulus enters the network and the counter that represents positivity is the first one in reaching the threshold (confusion may occur when a counter corresponding to a valence different from the stimulus reaches the threshold). For the lexical task, reaction times are computed in an analogous way, but in this case, nine counters are used