Summary: Learning new factual information about the world—such as details drawn from a fictional setting—engages brain regions that differ from those used to recall personal life events. In this study, participants learned made-up facts about imaginary civilizations and were tested later to see which facts they retained.
Functional brain scans showed that distinct cortical areas encoded semantic information about people and places during learning. Stronger representational activity in those regions predicted better memory for the facts, indicating that acquiring factual knowledge relies on a neural network that is at least partly separate from autobiographical memory systems.
Key Facts:
- Distinct Learning Systems: The brain recruits specialized regions for learning impersonal, semantic information that are different from those supporting personal episodic memories.
- Predictive Neural Signals: Greater activation and richer information patterns in regions representing people and places forecasted which facts participants later remembered.
- Fiction as a Research Tool: Using imaginary civilizations allowed researchers to control content and study how the brain encodes and stores new factual knowledge.
Source: SfN
Background: While prior research has mapped brain areas involved in remembering personal experiences, the neural bases that support learning impersonal, semantic facts about the world are less well understood.
In a new Journal of Neuroscience report, Scott Fairhall and colleagues at the University of Trento applied functional magnetic resonance imaging (fMRI) to study how humans acquire and later retrieve factual information. Twenty-nine volunteers completed a structured learning task in the scanner that presented fictional encyclopedic facts about people and places from three invented civilizations inspired by fantasy literature.
Participants studied 120 fictional facts and then returned for a memory test about a day and a half later. The delayed testing allowed researchers to compare which items were successfully encoded and later retrieved versus which were forgotten.
Analyses of the imaging data revealed multiple cortical regions sensitive to semantic content during encoding. Multivariate pattern analysis identified the medial precuneus, left angular gyrus, intraparietal sulcus, ventral occipitotemporal cortex, and lateral anterior temporal lobes bilaterally as areas where activity patterns reflected information about people and places.
Importantly, the representational strength of information within the precuneus and the left lateral anterior temporal lobe predicted whether a given fact would be remembered on the later test. In other words, the richness and distinctiveness of semantic representations in these regions during learning were linked with successful long-term retention.
The investigators did not find reliable differences using standard univariate approaches between remembered and forgotten items in many regions, suggesting that multivariate representational measures can reveal subtleties unobservable with simpler activation comparisons. Follow-up region-of-interest analyses indicated a possible role for response magnitude in the left inferior frontal gyrus, while medial temporal lobe structures showed neither univariate nor multivariate effects in this paradigm.
As Fairhall explains, these results support a model in which learning new factual knowledge is driven by the fidelity and richness of semantic representations in distributed cortical systems. That model is at least partly distinct from the mechanisms that support episodic memory, which typically rely on different neural substrates.
Key Questions Answered:
A: The study examined how the brain learns factual, impersonal information about the world, and whether distinct neural systems support that learning compared with autobiographical memory.
A: Volunteers learned fictional facts about people and places while undergoing fMRI; their memory for those facts was assessed roughly 1.5 days later to determine which items were retained.
A: Multivariate patterns in brain regions associated with semantic content—especially the precuneus and lateral anterior temporal lobe—predicted successful memory for newly learned facts, indicating distinct mechanisms for factual learning.
About this learning and memory research news
Author: SfN Media
Source: SfN
Contact: SfN Media – SfN
Image: The image is credited to Neuroscience News
Original Research: Closed access.
“Semantic Representational Strength in the Precuneus and Lateral ATL Predicts Successful Factual Learning” by Scott Fairhall et al., Journal of Neuroscience
Abstract
Semantic Representational Strength in the Precuneus and Lateral ATL Predicts Successful Factual Learning
Acquiring new factual knowledge is essential in everyday life. Although previous work has identified cortical regions involved in learning, the specific neural substrates that predict successful knowledge acquisition remained unclear. This fMRI study presented human participants (N=29; 7 male) with a naturalistic learning task composed of 120 fictional encyclopedic facts about people or places drawn from three invented civilizations.
Items that were later recalled were compared with those that were forgotten on an associative memory test administered about a day and a half after encoding. Multivariate pattern analysis complemented univariate approaches to reveal regions sensitive to semantic content during encoding: medial precuneus, left angular gyrus, intraparietal sulcus, ventral occipitotemporal cortex, and lateral anterior temporal lobes bilaterally.
Within these regions, the informational strength of representations in the precuneus and left lateral anterior temporal lobe predicted subsequent retrieval success. Planned analyses did not show reliable univariate differences between remembered and forgotten facts, highlighting the importance of representational measures. Targeted follow-up analyses suggested a possible role for left inferior frontal gyrus response magnitude, while medial temporal lobe structures did not show significant univariate or multivariate effects in this dataset.
Together, these findings support a model of factual learning that depends on the richness of semantic representations within distributed cortical systems, and that operates at least partially separately from episodic memory mechanisms.