Multi-Store Model (MSM)
Atkinson & Shiffrin (1968) proposed that memory consists of three distinct stores: sensory register, short-term memory (STM), and long-term memory (LTM). Information flows linearly through these stores via attention and rehearsal. The sensory register has a vast capacity but holds information for less than half a second. STM holds approximately 7±2 chunks (Miller, 1956) for about 18-30 seconds without rehearsal. LTM has potentially unlimited capacity and duration.
多存储模型
Atkinson和Shiffrin于1968年提出记忆由三个独立的存储系统组成：感觉登记器、短时记忆和长时记忆。信息通过注意力和复述在这些存储之间线性流动。感觉登记器容量巨大但信息保持不足半秒。短时记忆在无复述情况下可保持约7±2个组块（Miller, 1956），持续时间约18-30秒。长时记忆的容量和持续时间几乎是无限的。
The MSM is elegant but flawed. Its most glaring weakness is the assumption that information must pass through STM to reach LTM — a claim contradicted by Shallice and Warrington's (1970) case study of KF, a patient with impaired STM but intact LTM. This double dissociation suggests STM and LTM are not a single linear pipeline but operate in parallel to some degree. Furthermore, the MSM treats STM as a unitary, passive store, which the Working Memory Model directly challenges. MSM虽然简洁但存在明显缺陷。它最突出的问题是假设信息必须经过短时记忆才能进入长时记忆——这一说法被Shallice和Warrington于1970年对KF的案例研究所推翻。KF患者短时记忆受损但长时记忆完好。这种双重分离表明STM和LTM并非单一的线性管道，而是在一定程度上并行运作。此外，MSM将短时记忆视为一个单一被动的存储单元，工作记忆模型对此进行了直接挑战。Baddeley & Hitch (1974)
The Working Memory Model (WMM) replaces the passive STM with an active multi-component system. The central executive directs attention and coordinates three slave systems: the phonological loop (verbal/acoustic information), the visuospatial sketchpad (visual/spatial information), and the episodic buffer (added in 2000, integrates information across modalities and links to LTM). Dual-task experiments provide strong evidence: participants struggle to perform two tasks that compete for the same subsystem but can handle tasks using different subsystems simultaneously.
工作记忆模型
工作记忆模型用一个主动的多组件系统取代了被动的短时记忆。中央执行器负责分配注意力并协调三个子系统：语音回路（语言和声音信息）、视觉空间画板（视觉和空间信息）以及情景缓冲区（2000年新增，整合跨模态信息并与长时记忆连接）。双任务实验提供了强有力的证据：当两项任务竞争同一子系统时参与者表现困难，但使用不同子系统的任务可同时高效处理。
Students often confuse the phonological loop with the visuospatial sketchpad. Here is a simple heuristic: if a secondary task involves repeating "the the the" aloud (articulatory suppression), it occupies the phonological loop and disrupts verbal reasoning but leaves spatial reasoning intact. Conversely, tracking a moving dot of light occupies the visuospatial sketchpad and disrupts spatial tasks but spares verbal ones. The WMM explains phenomena the MSM cannot — including why we can simultaneously talk on the phone and navigate a familiar route, two tasks that would overload a unitary STM. 学生常混淆语音回路和视觉空间画板。一个简单的区分方法是：如果次要任务是重复默念"the the the"（发音抑制），它占用了语音回路，会干扰语言推理但空间推理不受影响。相反，跟踪移动的光点占用了视觉空间画板，会干扰空间任务但语言任务不受影响。WMM解释了MSM无法解释的现象——包括为什么我们可以一边打电话一边走熟悉的路线，这两项任务如果共用同一个短时记忆系统就会超负荷。 Beyond the structural models lies the question of depth: what makes some memories stick while others evaporate? Craik and Lockhart (1972) proposed the Levels of Processing (LOP) framework, arguing that memory is a byproduct of perceptual processing depth. Structural processing (shallow: "Is the word in capital letters?") produces the weakest trace. Phonemic processing (intermediate: "Does it rhyme with train?") yields moderate retention. Semantic processing (deep: "Would it fit in the sentence: The ___ galloped across the field?") produces the strongest and most durable memory trace. Elias et al. demonstrated this experimentally, showing that incidental learning under deep processing produced recall comparable to intentional learning under shallow processing — a finding with profound implications for study strategy. 在结构模型之外，还存在一个深度问题：为什么有些记忆能够巩固而有些则转瞬即逝？Craik和Lockhart于1972年提出了加工水平框架，认为记忆是知觉加工深度的副产品。结构加工（浅层："这个单词是大写吗？"）产生的记忆痕迹最弱。语音加工（中层："它和train押韵吗？"）产生中等保持力。语义加工（深层："它能否填入句子：___飞驰过田野？"）产生最强最持久的记忆痕迹。Elias等人通过实验证明，深层加工下的偶然学习产生的回忆效果与浅层加工下的有意学习相当——这一发现对学习策略有深远影响。Types of Long-Term Memory
Tulving (1972) distinguished three LTM systems. Episodic memory stores personally experienced events with their temporal and spatial context (your first day of school). Semantic memory stores facts and general knowledge divorced from personal context (Paris is the capital of France). Procedural memory stores motor skills and habits that operate largely outside conscious awareness (riding a bicycle). Brain imaging studies support this tripartite distinction: episodic memory activates the hippocampus and prefrontal cortex; semantic memory engages the temporal lobe; procedural memory relies on the cerebellum and basal ganglia.
长时记忆的类型
Tulving于1972年区分了三种长时记忆系统。情景记忆存储带有时间空间背景的个人经历事件（你上学的第一天）。语义记忆存储脱离个人背景的事实和一般知识（巴黎是法国首都）。程序性记忆存储运动技能和习惯，大部分在意识之外运作（骑自行车）。脑成像研究支持这种三元区分：情景记忆激活海马体和前额叶皮层；语义记忆涉及颞叶；程序性记忆依赖小脑和基底神经节。
Clinical evidence supports this tripartite distinction. Patient HM (Scoville & Milner, 1957) could not form new episodic or semantic memories after bilateral medial temporal lobectomy, yet could learn new motor skills — his procedural memory remained intact. This double dissociation confirms LTM is not a monolithic store but a federation of specialised systems. 临床证据支持这种三元区分。患者HM（Scoville & Milner, 1957）在接受双侧内侧颞叶切除术后无法形成新的情景或语义记忆，但仍可学习新的运动技能——程序性记忆保持完整。这种双重分离证明长时记忆并非单一存储，而是多个专门化系统的联盟。 Section two of the Edexcel specification pivots from how memory works to why it fails. Forgetting is not a design flaw but an adaptive feature of a cognitive system that must prioritise relevant information. Three major theories compete to explain the phenomenon. Edexcel大纲的第二部分从记忆如何运作转向记忆为何失效。遗忘不是认知系统的设计缺陷，而是其必须优先处理相关信息的一种适应性特征。三种主要理论竞争性地解释这一现象。 Interference theory holds that forgetting occurs not because memories decay but because similar memories compete and disrupt each other. McGeoch and McDonald (1931) demonstrated retroactive interference: participants who learned List A then List B recalled fewer List A items than a control group who rested between learning and recall. The degree of interference depends on similarity — learning a second list of synonyms produces more forgetting than learning nonsense syllables. Proactive interference works in the opposite direction: old learning disrupts new learning. Underwood (1957) found that participants who had learned multiple word lists in prior experiments showed dramatically poorer recall of a final list than naive participants. Interference theory has strong ecological validity: it explains why students who study French immediately before a Spanish exam may confuse vocabulary from the two languages. 干扰理论认为遗忘的发生不是因为记忆衰退，而是因为相似记忆之间的竞争和相互干扰。McGeoch和McDonald于1931年证明了倒摄抑制：学习列表A后再学习列表B的参与者，对列表A的回忆量少于学习与回忆之间休息的对照组。干扰程度取决于相似性——学习同义词列表比学习无意义音节产生的遗忘更多。前摄抑制方向相反：旧学习干扰新学习。Underwood于1957年发现，在先前实验中学习过多组词表的参与者对最后一组词表的回忆远低于初次参与者。干扰理论具有强大的生态效度：它解释了为什么在西班牙语考试前刚复习完法语的学生会混淆两种语言的词汇。 Retrieval failure theory offers a different explanation: memories are stored but temporarily inaccessible without the right cues. Tulving and Pearlstone (1966) demonstrated the encoding specificity principle: participants who learned words in the presence of category cues (e.g., "fruit — APPLE") recalled more when those same cues were provided at retrieval. This has real-world implications. Godden and Baddeley (1975) asked divers to learn word lists either on land or underwater. Recall was highest when the learning and retrieval contexts matched — the context-dependency effect. Similarly, Carter and Cassaday (1998) found that participants who learned material after taking an antihistamine (which causes mild drowsiness) recalled better when in the same drugged state — a state-dependency effect. 提取失败理论提供了不同解释：记忆被存储但缺乏正确线索时无法临时提取。Tulving和Pearlstone于1966年证明了编码特异性原则：在类别线索（如"水果——APPLE"）下学习单词的参与者在获得相同线索时回忆效果更好。这具有现实意义。Godden和Baddeley于1975年让潜水员在陆地或水下学习词表。当学习环境和提取环境匹配时回忆效果最高——这就是情境依赖效应。类似地，Carter和Cassaday于1998年发现服用抗组胺药（引起轻微嗜睡）后学习材料的参与者在相同药物状态下提取效果更好——状态依赖效应。 The third candidate, trace decay theory, is the most intuitive but the least supported. Hebb (1949) proposed that memory traces are physical changes in neural circuits that fade over time if not reactivated. Yet Peterson and Peterson's (1959) classic trigram experiment showed that decay alone cannot explain forgetting: participants' recall of consonant trigrams (e.g., "CJB") dropped from roughly 90% after 3 seconds to below 10% after 18 seconds. Crucially, the drop occurred only on later trials when proactive interference from earlier trigrams had accumulated — on the very first trial, recall was near-perfect even after 18 seconds. This suggests interference, not decay, is the primary mechanism. In the Edexcel exam, students should note that trace decay is difficult to test in isolation because the passage of time inevitably introduces opportunities for interference. 第三种理论即痕迹衰退理论，最为直观但支持最少。Hebb于1949年提出记忆痕迹是神经回路中的物理变化，若不重新激活会随时间消退。然而Peterson和Peterson于1959年的经典三元辅音实验表明仅凭衰退无法解释遗忘：参与者对辅音三元组（如CJB）的回忆从3秒后的约90%下降到18秒后的不足10%。关键在于，这种下降仅出现在后续试验中——当来自之前三元组的前摄抑制积累后。在第一次试验中，即使18秒后回忆仍近乎完美。这表明干扰而非衰退是主要机制。在Edexcel考试中，学生应注意痕迹衰退难以单独检测，因为时间流逝不可避免地引入了干扰的机会。
Eyewitness Testimony (EWT)
The third pillar of the Edexcel cognitive specification examines the reliability of eyewitness memory. Loftus and Palmer (1974) conducted the landmark study: participants watched a film of a car accident and were asked to estimate the cars' speed. The critical verb was varied — "smashed," "collided," "bumped," "hit," or "contacted." The "smashed" group estimated 40.8 mph on average; the "contacted" group, 31.8 mph. One week later, participants who heard "smashed" were more than twice as likely to falsely report seeing broken glass (which was not present in the film). This demonstrated the misinformation effect: post-event information can distort original memory traces.
目击证词
Edexcel认知心理学的第三个支柱考察目击记忆的可信度。Loftus和Palmer于1974年进行了标志性研究：参与者观看了一段车祸影片，然后被要求估算车速。关键动词被变换——"撞碎"、"碰撞"、"撞上"、"碰到"、"接触"。"撞碎"组平均估计40.8英里/小时；"接触"组平均31.8英里/小时。一周后，听到"撞碎"的参与者虚假报告看到碎玻璃（影片中并无碎玻璃）的概率是其他组的两倍多。这证明了误导信息效应：事后信息可以扭曲原始记忆痕迹。
The anxiety question complicates the EWT narrative. The Yerkes-Dodson law predicts an inverted-U relationship between arousal and performance: moderate anxiety improves memory, but extreme anxiety impairs it. Loftus (1979) proposed the weapon focus effect — when a weapon is present, witnesses fixate on it, narrowing attention and reducing recall of peripheral details including the perpetrator's face. However, Christianson and Hubinette (1993) studied real bank robbery witnesses and found that those directly threatened (high anxiety) actually recalled more detail than bystanders (moderate anxiety), challenging the inverted-U account. The Edexcel specification expects students to evaluate this contradiction: laboratory studies (high control, low ecological validity) tend to find weapon focus effects; field studies (low control, high ecological validity) sometimes show enhanced recall under high anxiety. Neither finding is definitive. 焦虑问题使目击证词的叙述更加复杂。Yerkes-Dodson定律预测唤醒与表现之间呈倒U型关系：中等焦虑改善记忆，但极端焦虑损害记忆。Loftus于1979年提出了武器聚焦效应——当武器在场时，目击者将注意力集中在武器上，缩小注意力范围并减少对外围细节（包括罪犯面孔）的回忆。然而Christianson和Hubinette于1993年研究了真实银行抢劫案的目击者，发现受到直接威胁的人（高焦虑）比旁观者（中等焦虑）回忆出更多细节，挑战了倒U型假说。Edexcel大纲要求学生评估这一矛盾：实验室研究（高控制力，低生态效度）倾向于发现武器聚焦效应；实地研究（低控制力，高生态效度）有时显示高焦虑下回忆增强。两者均非定论。The Cognitive Interview
Fisher and Geiselman (1992) developed the cognitive interview (CI) to maximise accurate eyewitness recall. It rests on four retrieval mnemonics. Mental reinstatement of context: the witness mentally reconstructs the physical and emotional environment of the event. Report everything: even details the witness deems trivial or incomplete are encouraged. Change order: recall the event in reverse or from a different starting point to disrupt schema-driven reconstruction. Change perspective: recall from another person's viewpoint to access different retrieval paths. A meta-analysis by Kohnken et al. (1999) of 55 studies found the CI produced 34% more correct information than standard police interviews, though it also increased incorrect details slightly — a trade-off students should address in evaluation.
认知访谈法
Fisher和Geiselman于1992年开发了认知访谈法以最大化目击者正确回忆。它基于四种提取策略。情境心理重建：目击者在心理上重构事件的物理和情感环境。报告一切：即使目击者认为琐碎或不完整的细节也鼓励报告。改变顺序：以倒序或不同起点回忆事件以打乱基于图式的重建。改变视角：从他人视角回忆以激活不同的提取路径。Kohnken等人于1999年对55项研究的元分析发现，认知访谈法比标准警察访谈多产生34%的正确信息，虽然错误细节也略有增加——学生在评估时应讨论这种取舍。
Schema theory, developed by Bartlett (1932) in his classic "War of the Ghosts" study, ties together memory reconstruction and eyewitness unreliability. Bartlett asked participants to read a Native American folk tale and reproduce it at increasing intervals. Recall became progressively shorter, more coherent, and more conventional — participants unconsciously altered details to fit their existing cultural schemas (mental frameworks). The unfamiliar element "something black came out of his mouth" was transformed into the more schema-congruent "he foamed at the mouth." Bartlett's conclusion — that memory is an active reconstruction shaped by prior knowledge, not a passive recording — remains foundational. For Edexcel students, schema theory provides a unified framework: it explains both the reconstructive nature of LTM (linking to Bartlett's own work) and the susceptibility of EWT to post-event misinformation (linking to Loftus). A schema is simultaneously a strength (it enables efficient processing) and a vulnerability (it introduces systematic distortion). 图式理论由Bartlett于1932年在其经典"幽灵之战"研究中提出，将记忆重建与目击者不可靠性联系起来。Bartlett让参与者阅读一则北美印第安民间故事，并在逐渐增加的时间间隔后复述。回忆内容变得越来越短、越来越连贯、越来越常规——参与者无意识地改变细节以适应当有文化图式（心理框架）。陌生的元素"有黑色的东西从他嘴里出来"被转换为更符合图式的"他口吐白沫"。Bartlett的结论——记忆是基于既有知识的主动重建而非被动记录——至今仍是基础性观点。对于Edexcel学生而言，图式理论提供了一个统一框架：它既解释了长时记忆的重建性（关联Bartlett自身的研究），也解释了目击证词对事后误导信息的敏感性（关联Loftus）。图式既是优势（使高效处理成为可能），也是弱点（引入系统性扭曲）。 For the Edexcel exam, students must move beyond describing studies to evaluating them using the GRAVE framework — Generalisability, Reliability, Application, Validity, and Ethics. Every study in this article can be evaluated through this lens, and top-band answers weave evaluation throughout rather than appending it as a final paragraph. 对于Edexcel考试，学生必须超越描述研究，使用GRAVE框架进行评估——概括性、信度、应用、效度和伦理。本文中每一项研究都可以通过这一视角进行评估，高分段答案应将评估贯穿全文而非作为最后一段附加。 A final note: the Edexcel Paper 1 16-mark essay demands AO1 (knowledge, 6 marks) and AO3 (evaluation, 10 marks). Describe two studies for AO1, then provide four well-developed evaluation points for AO3. Avoid spending too long on AO1 — the mark scheme rewards evaluation more generously. 最后提醒：Edexcel Paper 1的16分论述题中AO3评估占10分，AO1知识仅占6分。务必在评估上花费更多精力——评分方案对此赋分更慷慨。 🎓 想系统攻克Edexcel心理学Paper 1？需要一对一精讲记忆模型、干扰理论或目击证词的完整答题框架？扫码添加微信 16621398022，获取专属试听课和最新真题解析资料。16621398022同微信，备注"认知心理学"优先通过。更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图