Hippocampal System Dysfunction and Odor Discrimination Learning in Rats: Impairment or Facilitation Depending on Representational Demands

H. Eichenbaum, A. Pagan, and P. Mathews, N. J. Cohen

Behavioral Neurostience (1988)

DOI: 10.1037//0735-7044.102.3.331

The conceptual gap or the scientific question: if the representational demands were divergent, could the lesions of hippocampus produce distinct effects during the initial acquisition of each task?

How the authors proposed the question: it was a hypothesis-driven study. In one previous study, the authors found that the lesions of hippocampus facilitated odor-based reversal learning (shared yesterday, Normal Olfactory Discrimination Learning Set and Facilitation of Reversal Learning After Medial-Temporal Damage in Rats: Implications for an Account of Preserved Learning Abilities in Amnesia, JNS 1986). They proposed that the facilitation could be attributed to the switch, after hippocampus lesions, from the representations of relational information for different stimuli to the independent representation of each odor-outcome/stimulus-response associations. The relational information was redundant for animals to perform a go-nogo reversal discrimination task. To further verify this hypothesis, the authors designed three discrimination tasks, which imposed distinct representational demands, and compared the hippocampus-lesion effects. 

Brief summary: to verify their hypothesis (hippocampus encoding relational information), the investigators trained rats to learn three variants of odor-pair discrimination tasks: Simultaneous-Cue, Go-Left/Go-Right Task; Successive-Cue, Go-Left/Go-Right Task; Successive-Cue, Go/No-Go Task. They found that the lesions of hippocampus impaired, did not affect, and facilitated the performance of these three task, respectively. 

How the authors tell the story: in the Simultaneous-Cue, Go-Left/Go-Right Task, two odorants were delivered simultaneously from two odor ports with randomly associated spatial locations, which called for a direct comparison between stimuli, choices, and the corresponding outcomes. That is, a relational representation. In the Successive-Cue, Go/No-Go Task, only one odor was delivered in each trial, with one odor paired with reward in each odor pair. In this design, there was no requirement for direct comparison between stimuli, spatial-response information. So relational representation is not beneficial. The Successive-Cue, Go-Left/Go-Right Task have a mediocre requirement for relational representation: the comparison for spatial-response information, but not for sensory stimuli. 

Based on their hypothesis, they made specific predictions about the hippocampus-lesion effects on these three tasks: impairment for the Simultaneous-Cue, Go-Left/Go-Right Task; improvement for the Successive-Cue, Go/No-Go Task; intermediate effects for the Successive-Cue, Go-Left/Go-Right Task. And indeed, the experimental results exactly follow their predictions. So they wave a self-consistent and hypothesis-driven story.

The ongoing question(s) inspired by the current study: how training history affects the behavioral observations? The facilitation effect observed in the Successive-Cue, Go/No-Go Task was inconsistent with their JNS paper published in 1986 (shared yesterday). In that study, there was no significant effect on the first three go-nogo discrimination learning. On the contrary, the facilitation effect was only observed for the reversal learning after the third discrimination learning. I'm considering that whether it's a training history effect? Because in the present study, rats were trained to learn the Simultaneous-Cue, Go-Left/Go-Right Task, the Successive-Cue, Go-Left/Go-Right Task, and the Successive-Cue, Go/No-Go Task sequentially. The first two tasks have the representational requirement for relational information, which might teach animals to pay attention to the relations of sensory stimuli, choices, and outcomes. Such sort of relational representations were hindrance for animals to learn the Successive-Cue, Go/No-Go Task. So the lesions of fornix led to a facilitation effect.