There is a persistent metaphor that haunts memory work: the filing cabinet. The nervous system receives an experience, files it away in some neural folder, and retrieves it intact when needed. Memory is storage. Retrieval is finding what was stored.
But Kaufman and neuroscience converge on something stranger: storage is not filing. Every memory is scattered. It fragments across multiple neural systems simultaneously. The sensory cortex stores images and sounds separately from the amygdala's fear imprint, which is separate from the hippocampus's context, which is separate from the body's somatic record. These fragments do not communicate automatically. They do not arrive packaged.
Retrieval, therefore, is not finding. Retrieval is reconstruction. Every time you remember something, you are reassembling fragments that were never stored together, using present context and present affect to fill the gaps. The memory you retrieve is not the memory that was stored. It is a new construction built from old pieces.
This distinction—between storage as fragmentation and retrieval as reconstruction—changes everything about how we understand memory, trauma, healing, and the possibility of certainty itself.
The biological feed is the nervous system's encoding capacity. When an event occurs, multiple systems activate simultaneously. The sensory cortex captures images and sounds. The amygdala tags threat and affect. The hippocampus creates context and temporal sequence. The motor cortex and proprioceptive systems encode the body's response. The narrative brain (left hemisphere) begins constructing story. None of these systems speak the same language. None of them automatically integrate what the others recorded.
In a calm state, the prefrontal cortex can coordinate across these systems. Sensory data, affect, context, bodily sensation, and narrative can be woven into a coherent memory at the moment of encoding. But under threat—which is precisely when encoding matters most—the prefrontal cortex is offline. The amygdala is running. Coordinated encoding does not happen. The event is recorded in fragments.
This is not a failure. It is ancient design. The nervous system prioritizes survival in the moment over coherent memory. If a predator appears, your body does not wait for narrative integration. It responds. The memory fragments because the nervous system was doing what it was designed to do: keeping you alive in the present.
The systemic feed is the post-event environment and the availability of coherence. After the event, can the person talk about what happened? Can they be with someone who holds the experience with them? Can they move their body, discharge the affect, complete the interrupted response? In these conditions, the fragmented pieces begin to integrate into coherent narrative. The memory becomes a thing that happened—in the past, survivable, contextualized.
But if the person is alone, if they are shamed for what they're experiencing, if they cannot move their body or discharge the affect, the fragments stay scattered. They remain activated separately. A sound triggers the amygdala's fear response without the hippocampal context that says "that was then, this is now." An image activates without the narrative frame that says "I survived it." The fragments stay alive, reactivating independently.
Kaufman identifies what he calls "distinctive discontinuity" as the key to understanding how memories are actually stored.1 The memory is not stored as a unified narrative. It is stored as separate, unintegrated components that can be activated independently. The discontinuity is the gap between the fragments—the fact that they do not automatically cohere.
This means something counterintuitive: the more traumatic an event, the more fragmented the storage. A mildly distressing experience might be quickly integrated into narrative consciousness. But a severe threat creates such extensive fragmentation that the fragments may never spontaneously integrate. They remain scattered. They activate randomly. A person touches your arm and the amygdala fires—but not because the amygdala has a rational reason. It fires because a fragment of the memory (touch = danger) was stored without the context that would contradict it.
Retrieval, in this fragmented system, does not mean "pulling out the whole memory." It means activating one or more of the fragments. When you try to remember a traumatic event consciously, you are not retrieving the unified event. You are trying to activate fragments that were never unified. The conscious mind says "I want to remember." But which fragment will activate? The image? The sound? The fear? The shame? The bodily sensation? Whichever fragments activate are what "the memory" seems to be. But other fragments remain separate.
This explains one of the most disturbing facts about memory: certainty is not a measure of accuracy. A fragment that activates strongly—that carries intense affect or vivid sensation—feels absolutely real. The person is certain about it. But certainty is a feature of activation intensity, not historical accuracy. A fragment can be vividly, absolutely certain, and still be partially false, partially constructed, or entirely misremembered.
What the storage-retrieval mechanism reveals is that conscious awareness is always incomplete. The conscious mind has access only to the fragments that are currently activated. It does not have access to the full storage system. It does not know what is stored in the scattered pieces that are not activated. And it has no reliable way to distinguish between: (1) a fragment that is historically accurate, (2) a fragment that is partially accurate and partially constructed, (3) a fragment that is entirely false but felt with intense certainty.
This means that your sense of what happened—your conscious memory—is not a transcript of reality. It is an interpretation produced by whichever fragments your nervous system has activated, filtered through present affect, present context, and present meaning-making.
The implication is radical: you cannot trust your certainty. Not because you are lying or deluded, but because certainty is a property of activation, not of accuracy. You can be absolutely, viscerally certain about something that is partially or entirely false. And you can be uncertain about something that is completely true.
Retrieval also reveals something about meaning-making: humans cannot tolerate fragmentation. When fragments activate without context, the conscious mind automatically constructs narrative to hold them together. A person experiences fear and shame without understanding why. The mind does not sit with "I don't know why I feel this way." Instead, it reaches backward for a story. "I must have been abused. I must have been betrayed. I must have been inadequate." The story makes the feeling coherent. It transforms fragmentation into narrative. But the story may be constructed, not recovered.
Consider a woman in therapy who begins to "recover" memories of childhood abuse. The memories feel absolutely real. They are vivid, emotionally potent, accompanied by bodily sensations and affect. Over time, the memories grow more detailed and more severe. She becomes certain—absolutely certain—about what happened.
But when she confronts her family or attempts to verify the memories, the story collapses. Family members deny the events. No corroboration exists. Or worse: corroboration reveals a different story. Something did happen, but not in the form she remembered.
What happened? Three things simultaneously:
First, a fragment of real memory activated. Perhaps a genuine boundary violation. Perhaps legitimate discomfort or fear. A real piece of experience was stirred.
Second, the conscious mind, intolerant of fragmentation, began constructing narrative around the activated fragment. "This feeling means I was abused. This sensation means violation occurred. This shame means something terrible happened to me."
Third, through therapy and rehearsal, the constructed narrative became increasingly vivid and certain. With each telling, the story elaborated. New details emerged. The narrative felt more real because it had been rehearsed more. Repetition with emotional intensity created psychological reality.1 The constructed narrative became indistinguishable from recovered fact.
The woman is not lying. The therapist was not deliberately harmful. But a real fragment was magnified and supplemented with constructed material, and somewhere in the process, the boundary between recovery and construction disappeared.
If you are working to retrieve fragmented or uncertain memories:
Step 1 — Separate feeling from fact: The feeling is real. The affect is real. The somatic response is real. These require no verification. You experienced something. But the narrative explanation of what happened is a separate claim that may or may not be accurate.
Step 2 — Find the accessible fragment: What is the one piece of the memory that is already activated and certain? A sensation? A sound? An image? Do not force the whole memory to appear. Work with the one discontinuity point that is already live.
Step 3 — Activate the fragment safely: In a safe container—therapy, trusted relationship, somatic practice—deliberately activate that fragment. Let yourself feel it fully. Notice what emerges without forcing or suggesting.
Step 4 — Discover adjacent fragments: As one fragment activates, what other fragments emerge naturally? Do not imagine. Let the nervous system produce what is there. Notice whether these fragments cohere or remain separate.
Step 5 — Distinguish narrative construction from fragment activation: When your conscious mind creates a story to hold the fragments together, notice this as construction, not recovery. The construction serves a function—it makes the feeling bearable. But it is not necessarily historically accurate.
Step 6 — Accept uncertainty as the final state: You may never know with certainty what historically happened. This is tolerable. You can heal fragmented memory without settling the historical fact. Healing happens through integration—moving from scattered activation to coherent nervous system response—not through historical certainty.
Memory retrieval fails when:
When retrieval fails, the person may remain symptomatic—the fragments continue to activate involuntarily, continuing to distress the nervous system. Different therapeutic approaches may be needed. Longer-term safety building. Somatic work that does not require narrative. Recognition that integration may not be achievable through the specific memory approach being attempted.
Evidence: Kaufman's reconstruction model is supported by neuroscience. Fragmented encoding is documented. The role of distinctive discontinuity in memory retrieval is clinically observable. Studies show that certainty about a memory does not correlate with historical accuracy. Repeated retrieval and rehearsal increase confidence in memories but do not increase accuracy—they increase construction.1
Tensions: Yet tension remains. Some memories are genuinely accurate despite uncertainty. Some retrieved memories are historically true. Some people recover genuine abuse through therapeutic memory work. Not all reconstruction is false. How do we distinguish between legitimate recovery and false construction? We cannot use certainty as the measure. We cannot use vividness. The mechanisms appear identical.
Open Questions:
Kaufman's reconstruction model of memory stands in productive tension with recovered memory advocates (who assume all integrated memories are real) and memory skeptics (who dismiss many recovered memories as false). Kaufman's position is more nuanced: reconstruction is a universal mechanism. Trauma is real. Memory fragmentation is real. But the relationship between historical events, nervous system encoding, and conscious narrative is complex. A memory can be psychologically real without being historically accurate. Emotional truth and factual truth are not the same dimension.
This contrasts with approaches that treat memory as either a faithful recording or a reliable reconstruction system. Kaufman reveals that neither is true—memory is reconstruction that can be psychologically and even therapeutically valid while being historically uncertain. The healing happens through the reconstruction process itself, not through achieving historical certainty.
Where neuroscience reveals that memory is stored across multiple systems that do not automatically integrate—sensory cortex, amygdala, hippocampus, motor cortex, all encoding the same event in incompatible formats—psychology describes retrieval as if a unified memory exists to be retrieved. The tension reveals that consciousness creates an illusion of unified memory by assembling fragments on demand. The conscious mind does not perceive fragmentation because it only accesses the activated fragments. It experiences retrieval as "remembering the event" when it is actually reconstructing it from scattered pieces.
This insight that neither domain generates alone: the experience of unified memory is a cognitive construction, not a neural reality. Your consciousness stitches together fragments that were never unified, using present context and present affect to fill the gaps, and experiences the stitching as remembering. This is why certainty about a memory feels identical whether the memory is accurate or constructed—because the mechanism is identical. The conscious experience of remembering is the same whether you are accessing a genuinely stored fragment or unconsciously constructing a plausible narrative. The nervous system has no epistemic marker that distinguishes them. Only external verification can reveal the difference.
The therapeutic implication that follows: healing memory does not require resolving this ambiguity. Healing requires integration—moving from fragmented activation to coherent nervous system response. A person can achieve full nervous system integration (and symptom relief) through activating and reconstructing a memory that may be entirely false. The healing is real. The truth is uncertain. These are not incompatible—they operate at different levels. The nervous system cares about coherence; courts care about facts. Both are real demands, and they cannot be met simultaneously.
Where epistemology distinguishes between empirical truth (what happened?), logical truth (what must be?), and phenomenological truth (what is the experience?), memory collapses all three. A historical fact (empirical) may be inaccessible. The psychological reality (phenomenological) may be the only available truth. Yet law, history, and therapy all have stakes in memory—but they cannot use the same standard of truth.
The tension reveals something fundamental about knowledge itself: memory cannot be separated from the act of remembering. The stored fact (if it exists) is not accessible independently of the retrieval process. And the retrieval process is inherently reconstructive. Therefore, we cannot have unmediated access to historical truth through memory. We have only the memory as it is reconstructed—which is always a hybrid of original encoding, present context, and present meaning-making.
This does not mean memory is useless as evidence. It means we must stop expecting memory to provide the kind of certainty that epistemology requires. What memory provides is phenomenological truth—the person's actual lived experience. That experience may not align with historical fact. But the experience is real and has shaped the person's nervous system. Integration of the experience (whether historically accurate or not) can heal the nervous system's dysregulation.
The deeper insight: different domains require different truth standards because different domains serve different functions. History needs facts to preserve what happened. Psychology needs coherence to restore nervous system regulation. Law needs certainty to establish liability. Epistemology needs rigor to distinguish true from false. Memory cannot satisfy all these standards simultaneously. The person becomes trapped when they demand that one memory serve all these functions at once.
Your most certain memories may be your most constructed ones. If you have a vivid, emotionally potent memory of abuse, that memory may be real, may be partially real, or may be constructed. The vividness is not an indicator. The emotional intensity is not an indicator. Your certainty is absolutely not an indicator. You can feel completely certain about something that is historically false. This means you may never know with certainty what happened. And you may not need to know. You can heal the nervous system response without settling whether the historical facts are accurate. The fragmentation itself—not the truth or falsity of the recovered narrative—is what damages the nervous system. Integration heals it, whether the narrative is true or false.
Question 1: If historical accuracy is not necessary for healing, what is the difference between memory recovery and imagination? Why does the form of the narrative matter if only the integration changes the nervous system response?
Question 2: A person with uncertain memories can heal. But can they also gain genuine knowledge about what happened to them? Is there a way to distinguish legitimate recovery from false construction while healing simultaneously, or are these mutually exclusive goals?
Question 3: Reconstruction is universal and inevitable. But does that mean we should stop trying to distinguish accurate memories from constructed ones? Or is the epistemic effort itself valuable even if certainty is impossible?