Untitled

Squire 2009 — Memory and Brain Systems: 1969–2009

Reference: Larry R. Squire, Journal of Neuroscience 29(41), 12711–12716. DOI, public full text. Type: invited historical review, not a new experiment. Read on 22 September 2026.

Reading record. Complete main article read from the saved PMC HTML/text, including its discussion and closing qualifications. The cited original studies were not all read. No supplementary experiment or analysis code is associated with this reading. A PDF request returned an access-check page; the valid article HTML and text are retained instead, with provenance alongside them.

Source account, in brief. Squire organizes memory into interacting systems rather than one store. Declarative recollection, skills/habits, conditioning, priming, and working-memory control have distinguishable behavioral and anatomical evidence. The same experience can yield reportable content and nondeclarative effects. The review favors a temporally limited hippocampal contribution to long-term storage while acknowledging dispute over spatial and autobiographical memory. It distinguishes memory content from strategic retrieval and source monitoring. It also discusses sleep, reinstatement, and retrieval-related modification, noting unresolved mechanisms. These are the author's synthesis of the literature available in 2009, not a current consensus verdict or one dataset establishing every claim.

Our interpretation for the research program.

The extraction target should be specified by the effect we hope to preserve. An interview can obtain someone's description of an experience without identifying the habits, expectations, or response dispositions associated with it. Conversely, a behavioral tendency could be measurable even when no episode is reported. A single text field labeled “memory” would hide these different observation problems.

This motivates a proposed functional inventory with separate entries for acquired content, cue-dependent accessibility, learned response tendencies, and strategic use. It does not require copying the review's anatomical taxonomy into software. A simpler representation may reproduce the relevant functions; that is an experimental question.

An especially useful dissociation for later experiments is knowledge versus use: give several candidate models the same information, then vary cues, source-monitoring demands, temporal-order questions, and future intentions. Differences in performance would locate a functional shortcoming more precisely than additional factual recall tests. Each task still needs ordinary cognitive and task-specific baselines.

What this reading does not settle. The review cannot establish a decoder from neural structure, a sufficient set of observations for a person, or the necessity of any particular artificial architecture. Its strong historical framing of consolidation should be compared with competing accounts and subsequent primary evidence. Follow-up priorities are human lesion dissociations, hippocampal indexing/pattern completion, and standard consolidation versus multiple-trace/transformation theories. Those disagreements should be evaluated at the level of what is retained and what the task asks, rather than resolved by choosing a preferred review.