海馬在情境恐懼制約中的角色：單純聯結之外

Abstract

組態學習是將多重刺激結合成一體的表徵歷程，此學習被認為是由海馬所主導。動物研究中常使用情境恐懼制約作業探討組態學習，過去研究咸信此作業涉及二個學習歷程，動物必須先將多向度的環境刺激整合成為單一的情境表徵，再將此表徵與電擊聯結。雖然過去許多研究認為在情境恐懼制約中，海馬參與情境表徵，而杏仁核則參與表徵和電擊的聯結。根據海馬整合多相度刺激的功能，本論文探討海馬在情境恐懼制約的角色是否也涉及融合電擊與情境訊息成一整體的恐懼情境表徵。本論文採用二階段訓練的策略來檢驗這樣的想法，大白鼠先於第一天情境探索形成情境表徵，第二天再於該情境中接受電擊。在這兩個訓練階段分別於海馬內注射局部麻醉劑二丁卡因壓抑其功能，從第三天記憶測試的表現便可得知海馬在兩學習歷程中分別具有的功能。結果發現於兩階段訓練歷程前後抑制背側海馬均會損害情境制約的僵懼反應，但同樣的實驗操弄施於腹側海馬則無效。另外，於情境-電擊習得階段於背側海馬施予氮甲基天門冬氨酸受體拮抗劑APV，也會造成記憶損害，這表示背側海馬除了參與情境表徵的建立之外，也負責情境-電擊學習。為了檢驗海馬於情境-電擊學習階段所負責的記憶是否有別於杏仁核所負責的簡單聯結記憶，本論文進一步證實情境-電擊階段的學習具有組態學習的彈性提取的特徵，可以在短暫呈現電擊或相關提取線索後，於非制約情境的測試箱中而表現出僵懼反應。更值得注意的是，於情境-電擊學習階段暫時抑制背側海馬會使得此記憶表現受損。相對的，抑制杏仁核則沒有效果。這表示背側海馬與杏仁核在情境-電擊階段所負責的記憶功能可能不同。本研究進一步發現，於學習後於背側海馬施予伽馬-胺基丁酸A類受體促進劑muscimol會影響情境表徵的建立，但不影響恐懼情境表徵的形成。施予蘑菇膽鹼受體拮抗劑scopolamine則只影響恐懼情境表徵的形成，不影響情境表徵的建立。以上結果支持，在情境恐懼制約中，動物除了可能將情境與電擊形成簡單聯結之外，也可能透過海馬將情境與電擊整合成為一體的情境表徵。而恐懼情境學習所涉及的神經機制與情境學習不同可能反映了兩種不同歷程的組態學習。情境表徵的建立倚賴靜態環境中，刺激的穩定關係，而恐懼經驗的組態表徵則是針對動態刺激的出現，修正原有的情境表徵。本研究可以提供未來在焦慮疾患的治療理論基礎與人類記憶系統研究的動物模型。

Configural learning refers to a process integrating elemental features into a unitary representation, which is often conceived to rely on the hippocampus. In past years, contextual fear conditioning in rodents has often been used to study neural mechanisms underlying configural learning. Many researchers conceive that contextual fear conditioning involves two learning processes: binding multiple cues in the context to form a unitary representation relying on the hippocampus, and then associating this representation to a shock relying on the amygdala. In light of evidence that the hippocampus is involved in binding multiple stimuli temporally or spatially related, it is proposed in this dissertation that the hippocampus is engaged not only in coding contextual information but also in integrating context and shock into a configural fear memory. To test this hypothesis, this dissertation adopted a two-phase training paradigm and examined the effects of manipulating the hippocampus at each phase independently. Suppressing the dorsal hippocampus (DH), but not its ventral part, by infusing 4% lidocaine during each training phase impaired conditioned freezing. Intra-DH infusion of DL-2-amino-5-phosphonovaleric acid, a N-methyl-D-aspartic acid antagonist prior to context-shock learning yielded the same results, suggesting that the DH is involved in both context coding as well as context-shock learning. Results further showed that context-shock training forges in the DH a new representation of frightful context representation by integrating environmental and shock cues altogether, which was readily retrieved through direct or mediated pattern completion, a kernel property of configural representation. Finally, intra-DH infusion of the GABAA receptor agonist muscimol immediately after context coding impaired contextual fear conditioning but had no effect if given after context-shock leaning. In contrast, intra-DH infusion of the muscarinic receptor antagonist scopolamine impaired contextual fear conditioning only if given immediately after context-shock learning but not given after context coding. These findings, taking together, suggest that in contextual fear conditioning the DH through different types of neurochemical operation may not only incorporate static environmental cues to form a configural memory for a neutral context but also incorporate shock-generated cues into a neutral context and transform it into a configural fear memory. These findings provide theoretical implications for studying underlying mechanisms of human learning and memory.