Multiple Trace Theory of Memory

Multiple Trace Theory

Multiple trace theory is a memory consolidation model developed as an alternative to strength theory. It holds that each time information is presented to a person, it is neurally encoded in a distinct memory trace made up of a mixture of its properties.

This idea diverges from the once-dominant standard model of memory consolidation, which suggested that the hippocampus was primarily involved only in the initial storing of memories, which were then transferred to the cortex for long-term storage. This approach makes it easier to explain the very long gradients that are common in retrograde amnesia.

Central to Multiple Trace Theory (MTT) are a few fundamental principles:

  • Memory is distributed: Each time an event is remembered, the hippocampus generates a distinct but overlapping neurological pattern or trace.
  • Engram complexity: An engram—the physical embodiment of a memory—becomes more complex with each recall, as new details may integrate with existing ones.
  • Persistence of the hippocampus: Contrary to the Standard Model, MTT maintains that the hippocampus remains involved in memory retrieval, even for old memories.

Comparing Multiple Trace Theory with the Standard Model, one finds key differences:

  • Hippocampal involvement: The Standard Model argues for a time-dependent shift in reliance from the hippocampus to the cortex, whereas MTT underscores continuous hippocampal engagement for memory retrieval.
  • Memory trace alteration: Unlike the Standard Model, which intuits that a memory trace, once consolidated, remains static, MTT suggests that these traces can be updated or altered upon each recall.
  • Memory representation: While the Standard Model leans towards a more unitary representation of memories, MTT supports the notion that memories can exist in multiple versions, represented by unique traces in the brain.

Neurobiological Underpinnings

Engram cells are the fundamental units in the neurobiology of memory, operating as tangible correlates of memory. Research has indicated that these cells get activated during learning. They undergo changes such as synaptic strengthening, which is pivotal for engram formation and subsequent recall of information.

The hippocampus plays a crucial role in consolidating information from short-term to long-term memory, indicating its importance in the context of MTT. Studies have shown that hippocampal lesions can severely disrupt this process, leading to difficulties in forming new memories, highlighting the hippocampus’s critical function in memory encoding and retrieval.

Within the medial temporal lobe, which houses the hippocampus, alterations in neural circuitry can lead to changes in how memories are processed and preserved. The medial temporal lobe is integral for memory formation, and its integrity is essential for the proper functioning of memory systems. Damage to this region, such as through lesions, can result in profound memory impairments.

MTT Attributes

The characteristics of an item determine its trace and can be classified into a variety of categories. When an item is committed to memory, information from each of these attributional categories is stored in the object’s trace. There may be a type of semantic classification at work, in which an individual trace is absorbed into larger conceptions about an object.

For example, when a person sees a duck, a trace is added to the “duck” cluster of traces in his or her brain. This new “duck” trace, while unique and distinct from other ducks seen in the person’s life, serves to reinforce the more general and overarching concept of a duck.

Physical attributes of an item encode information about the physical properties of the item being presented. For a word, this could include color, font, spelling, and size, but for an image, it could contain object forms and colors.

Experiments have demonstrated that persons who are unable to recall an individual word can sometimes recall the first or final letter, as well as rhyming syllables, all of which are contained in the physical orthography of a word’s trace. Even if an item is not visually exhibited, when encoded, it may have physical properties that are dependent on a visual representation.

Modality attributes contain information about the method by which an item was presented. The most common forms of modalities in an experimental context are auditory and visual. Any sensory modality can be used practically.


Contextual attributes are a broad class of properties that describe the item’s internal and external features while it is being presented. Internal context refers to the internal network that a trace elicits.

This can include features of an individual’s mood as well as other semantic associations evoked by the word’s presentation. External context, on the other hand, encodes geographical and temporal information as it is provided. This could represent the time of day or the weather, for instance.

Spatial qualities can refer to both the physical and imagined world. The loci technique, a mnemonic strategy that incorporates an imagined physical position, allocates relative spatial positions to various recalled elements before “walking through” these assigned positions to remember the contents.

Memory Types and Multiple Trace Theory

This theory applies differently to varied forms of memory, most notably episodic, semantic, and autobiographical memories, which are processed and consolidated in the brain through potentially different mechanisms.

Episodic memory refers to the capacity to recall specific events from one’s own past, often tagged with particular times and places. MTT posits that every time an episodic memory is retrieved, it is altered, potentially creating multiple traces in the hippocampus (HPC).

This contrasts with semantic memory, which encompasses knowledge about the world that is devoid of personal context, such as facts and concepts. Unlike episodic memories, semantic memories may become less dependent on the hippocampus over time, leading to debate on whether they also follow a multiple trace pattern.

Research into autobiographical memory — memories of one’s own life history — has further illuminated MTT’s implications. Studies have suggested that over time, autobiographical memories might integrate episodic and semantic details, thereby reinforcing the theory that multifaceted memories can generate several traces within the memory system.

The MTT framework has been notably supported by computational, neuroimaging, and neuropsychological results, indicating that memories may not reside in a single location but are distributed across different brain regions that work cooperatively, a concept aligned with the complementary learning systems theory.


Multiple trace theory fits nicely within the conceptual framework for recognition. A person must identify whether or not they have previously seen an item in order to recognize it. For example, facial recognition determines whether or not a person has previously seen a particular face.

When asked for a successfully encoded item (something that has previously been viewed), recognition should occur with high probability. In the mathematical framework of this theory, we can model recognition of an individual probe item p as accumulated similarity to a criterion. We convert the test item into an attribute vector, as we did with encoded memories, and compare it to every trace we’ve ever encountered.

If the aggregate similarity meets the condition, we declare we’ve seen the item previously. The summed similarity is expected to be very low if the item has never been seen, but somewhat high if it has, due to the probe’s properties being similar to some memory in the memory matrix.

Memory phenomena such as repetition, word frequency, recency, forgetting, and contiguity, among others, can be straightforwardly explained in terms of multiple trace theory. Memory has been shown to improve with repeated exposure to items.

For example, hearing a word numerous times in a list improves its identification and recall later on. This is because repeated exposure just adds the memory to the ever-growing memory matrix, resulting in a bigger summed similarity for this memory, making it more likely to meet the requirement.

Recency in the serial position effect can be explained by the fact that more recent memories encoded will have a temporal context that is most similar to the current situation, since time’s stochastic character will not have had as strong an effect.

As a result, context similarity will be high for newly encoded items, implying that overall similarity will also be high. The stochastic contextual drift is also assumed to account for forgetting because, as the context in which a memory was encoded fades over time, the cumulative similarity for an object only presented in that context decreases.


One of the most significant drawbacks of multiple trace theory is the need for some item to compare the memory matrix to in order to determine successful encoding. This works well for recognition and cued recall, but there is a clear difficulty to incorporate free recall into the model.

Free recall necessitates the ability to recall a list of items at will. Although the act of asking to recall may serve as a trigger to evoke cued recall procedures, the cue is unlikely to be distinct enough to meet a cumulative similarity threshold or obtain a high probability of recall.

Another key challenge is translating the model to biological relevance. It’s difficult to believe that the brain has an infinite ability to keep track of such a massive matrix of memories and continue to expand it with each new item presented. Furthermore, navigating through this matrix is an exhausting operation that has little relevance on biological time scales.

  1. Brown, R; Mcneil D (1966). The “tip of the tongue” phenomenon. Journal of Verbal Learning and Verbal Behavior. 5 (4): 325–337. doi: 10.1016/s0022-5371(66)80040-3
  2. Hintzman, DL; Block RA (1971). Repetition and memory: Evidence for a multiple trace hypothesis. Journal of Experimental Psychology. 88 (3): 297–306. doi: 10.1037/h0030907
  3. Karpicke, JD; Roediger HL (2007). Repeated retrieval during learning is the key to long-term retention. Journal of Memory and Language. 57: 151–162. doi:10.1016/j.jml.2006.09.004
  4. Nadel, L., Samsonovich, A., Ryan, L., & Moscovitch, M. (2000). Multiple trace theory of human memory: computational, neuroimaging, and neuropsychological results. Hippocampus, 10(4), 352-368.
  5. Ortega-de San Luis, C., & Ryan, T. J. (2022). Understanding the physical basis of memory: Molecular mechanisms of the engram. The Journal of biological chemistry, 298(5), 101866.
  6. Sutherland, R. J., Lee, J. Q., McDonald, R. J., & Lehmann, H. (2020). Has multiple trace theory been refuted? Hippocampus, 30(8), 842–850.