Common and distinct correlates of construction and elaboration of episodic-autobiographical memory: An ALE meta-analysis

The memory systems of the human brain and generative artificial intelligence

Generative Artificial Intelligence foundation models (for example Generative Pre-trained Transformer - GPT - models) can generate the next token given a sequence of tokens. How can this 'generative AI' be compared with the 'real' intelligence of the human brain, when for example a human generates a whole memory in response to an incomplete retrieval cue, and then generates further prospective thoughts? Here these two types of generative intelligence, artificial in machines and real in the human brain are compared, and it is shown how when whole memories are generated by hippocampal recall in response to an incomplete retrieval cue, what the human brain computes, and how it computes it, are very different from generative AI. Key differences are the use of local associative learning rules in the hippocampal memory system, and of non-local backpropagation of error learning in AI. Indeed, it is argued that the whole operation of the human brain is performed computationally very differently to what is implemented in generative AI. Moreover, it is emphasized that the primate including human hippocampal system includes computations about spatial view and where objects and people are in scenes, whereas in rodents the emphasis is on place cells and path integration by movements between places. This comparison with generative memory and processing in the human brain has interesting implications for the further development of generative AI and for neuroscience research.

The formation of episodic autobiographical memory is predicted by mental imagery, self-reference, and anticipated details

Introduction

Despite the ecological nature of episodic memory (EM) and the importance of consolidation in its functioning, studies tackling both subjects are still scarce. Therefore, the present study aims at establishing predictions of the future of newly encoded information in EM in an ecological paradigm.

Methods

Participants recorded two personal events per day with a SenseCam portable camera, for 10 days, and characterized the events with different subjective scales (emotional valence and intensity, self-concept and self-relevance, perspective and anticipated details at a month, mental images…). They then performed a surprise free recall at 5 days and 1 month after encoding. Machine learning algorithms were used to predict the future of events (episodic or forgotten) in memory at 1 month.

Results

The best algorithm showed an accuracy of 78%, suggesting that such a prediction is reliably possible. Variables that best differentiated between episodic and forgotten memories at 1 month were mental imagery, self-reference, and prospection (anticipated details) at encoding and the first free recall.

Discussion

These results may establish the basis for the development of episodic autobiographical memory during daily experiences.

Neural substrates of specific and general autobiographical memory retrieval in younger and older adults

Healthy aging is associated with a shift away from the retrieval of specific episodic autobiographical memories (AMs), towards more general and semanticized memories. Younger adults modulate activity in the default mode network according to the episodic specificity of AM retrieval. However, little is known about whether aging disrupts this neural modulation. In the current study we examine age-related changes in the modulation of whole-brain networks in response to three tasks falling along a gradient of episodic specificity. Younger and older adults retrieved specific (unique) AMs, general (routine) AMs, and semantic (general knowledge) memories. We found that both younger and older adults modulated default mode regions in response to varying episodic specificity. In addition, younger adults upregulated activity in several default mode regions with increasing episodic specificity, while older adults either did not modulate these regions, or downregulated activity in these regions. In contrast, older adults upregulated activity in the left temporal pole for tasks with higher episodic specificity. These brain activation patterns converge with prior findings that specific AMs are diminished in episodic richness with age, but are supplemented with conceptual and general information. Age-related reductions in the modulation of default mode regions might contribute to the shift away from episodic retrieval and towards semantic retrieval, resulting in reduced episodic specificity of personal memories.