Pub Date : 2024-05-24DOI: 10.1038/s41562-024-01897-6
Julia P. G. Jones, Ganga Shreedhar
Causal inference is needed to understand whether conservation is working. There is a substantial role for behavioural science, as interventions often depend on behaviour change. A focus on design over data, embracing mixed methods and support from funders will help to provide the evidence needed to reverse biodiversity loss.
{"title":"The causal revolution in biodiversity conservation","authors":"Julia P. G. Jones, Ganga Shreedhar","doi":"10.1038/s41562-024-01897-6","DOIUrl":"10.1038/s41562-024-01897-6","url":null,"abstract":"Causal inference is needed to understand whether conservation is working. There is a substantial role for behavioural science, as interventions often depend on behaviour change. A focus on design over data, embracing mixed methods and support from funders will help to provide the evidence needed to reverse biodiversity loss.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141091844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-20DOI: 10.1038/s41562-024-01882-z
James W. A. Strachan, Dalila Albergo, Giulia Borghini, Oriana Pansardi, Eugenio Scaliti, Saurabh Gupta, Krati Saxena, Alessandro Rufo, Stefano Panzeri, Guido Manzi, Michael S. A. Graziano, Cristina Becchio
At the core of what defines us as humans is the concept of theory of mind: the ability to track other people’s mental states. The recent development of large language models (LLMs) such as ChatGPT has led to intense debate about the possibility that these models exhibit behaviour that is indistinguishable from human behaviour in theory of mind tasks. Here we compare human and LLM performance on a comprehensive battery of measurements that aim to measure different theory of mind abilities, from understanding false beliefs to interpreting indirect requests and recognizing irony and faux pas. We tested two families of LLMs (GPT and LLaMA2) repeatedly against these measures and compared their performance with those from a sample of 1,907 human participants. Across the battery of theory of mind tests, we found that GPT-4 models performed at, or even sometimes above, human levels at identifying indirect requests, false beliefs and misdirection, but struggled with detecting faux pas. Faux pas, however, was the only test where LLaMA2 outperformed humans. Follow-up manipulations of the belief likelihood revealed that the superiority of LLaMA2 was illusory, possibly reflecting a bias towards attributing ignorance. By contrast, the poor performance of GPT originated from a hyperconservative approach towards committing to conclusions rather than from a genuine failure of inference. These findings not only demonstrate that LLMs exhibit behaviour that is consistent with the outputs of mentalistic inference in humans but also highlight the importance of systematic testing to ensure a non-superficial comparison between human and artificial intelligences. Testing two families of large language models (LLMs) (GPT and LLaMA2) on a battery of measurements spanning different theory of mind abilities, Strachan et al. find that the performance of LLMs can mirror that of humans on most of these tasks. The authors explored potential reasons for this.
{"title":"Testing theory of mind in large language models and humans","authors":"James W. A. Strachan, Dalila Albergo, Giulia Borghini, Oriana Pansardi, Eugenio Scaliti, Saurabh Gupta, Krati Saxena, Alessandro Rufo, Stefano Panzeri, Guido Manzi, Michael S. A. Graziano, Cristina Becchio","doi":"10.1038/s41562-024-01882-z","DOIUrl":"10.1038/s41562-024-01882-z","url":null,"abstract":"At the core of what defines us as humans is the concept of theory of mind: the ability to track other people’s mental states. The recent development of large language models (LLMs) such as ChatGPT has led to intense debate about the possibility that these models exhibit behaviour that is indistinguishable from human behaviour in theory of mind tasks. Here we compare human and LLM performance on a comprehensive battery of measurements that aim to measure different theory of mind abilities, from understanding false beliefs to interpreting indirect requests and recognizing irony and faux pas. We tested two families of LLMs (GPT and LLaMA2) repeatedly against these measures and compared their performance with those from a sample of 1,907 human participants. Across the battery of theory of mind tests, we found that GPT-4 models performed at, or even sometimes above, human levels at identifying indirect requests, false beliefs and misdirection, but struggled with detecting faux pas. Faux pas, however, was the only test where LLaMA2 outperformed humans. Follow-up manipulations of the belief likelihood revealed that the superiority of LLaMA2 was illusory, possibly reflecting a bias towards attributing ignorance. By contrast, the poor performance of GPT originated from a hyperconservative approach towards committing to conclusions rather than from a genuine failure of inference. These findings not only demonstrate that LLMs exhibit behaviour that is consistent with the outputs of mentalistic inference in humans but also highlight the importance of systematic testing to ensure a non-superficial comparison between human and artificial intelligences. Testing two families of large language models (LLMs) (GPT and LLaMA2) on a battery of measurements spanning different theory of mind abilities, Strachan et al. find that the performance of LLMs can mirror that of humans on most of these tasks. The authors explored potential reasons for this.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11272575/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141071129","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-15DOI: 10.1038/s41562-024-01892-x
Mike D. Schneider, Temitope O. Sogbanmu, Hannah Rubin, Alejandro Bortolus, Emelda E. Chukwu, Remco Heesen, Chad L. Hewitt, Ricardo Kaufer, Hanna Metzen, Veli Mitova, Anne Schwenkenbecher, Evangelina Schwindt, Helena Slanickova, Katie Woolaston, Li-an Yu
{"title":"Science–policy research collaborations need philosophers","authors":"Mike D. Schneider, Temitope O. Sogbanmu, Hannah Rubin, Alejandro Bortolus, Emelda E. Chukwu, Remco Heesen, Chad L. Hewitt, Ricardo Kaufer, Hanna Metzen, Veli Mitova, Anne Schwenkenbecher, Evangelina Schwindt, Helena Slanickova, Katie Woolaston, Li-an Yu","doi":"10.1038/s41562-024-01892-x","DOIUrl":"10.1038/s41562-024-01892-x","url":null,"abstract":"","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140925125","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01877-w
Humans are unusually adept at endurance running, due in part to specialized muscle fibres and heat elimination by sweating. Cost–benefit analyses and an ethnohistorical survey of hunting methods suggest that these features could have evolved through the pursuit of evasive species until they are overcome with exhaustion and easily dispatched.
{"title":"Ethnohistorical analysis suggests that endurance running evolved with persistence hunting","authors":"","doi":"10.1038/s41562-024-01877-w","DOIUrl":"10.1038/s41562-024-01877-w","url":null,"abstract":"Humans are unusually adept at endurance running, due in part to specialized muscle fibres and heat elimination by sweating. Cost–benefit analyses and an ethnohistorical survey of hunting methods suggest that these features could have evolved through the pursuit of evasive species until they are overcome with exhaustion and easily dispatched.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140915094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01886-9
Chengkun Yang, Xiaoxi Zhang, Shi Yan, Sizhe Yang, Baihui Wu, Fengshuo You, Yue Cui, Ni Xie, Zhiyi Wang, Li Jin, Shuhua Xu, Menghan Zhang
The Han Chinese history is shaped by substantial demographic activities and sociocultural transmissions. However, it remains challenging to assess the contributions of demic and cultural diffusion to Han culture and language, primarily due to the lack of rigorous examination of genetic–linguistic congruence. Here we digitized a large-scale linguistic inventory comprising 1,018 lexical traits across 926 dialect varieties. Using phylogenetic analysis and admixture inference, we revealed a north–south gradient of lexical differences that probably resulted from historical migrations. Furthermore, we quantified extensive horizontal language transfers and pinpointed central China as a dialectal melting pot. Integrating genetic data from 30,408 Han Chinese individuals, we compared the lexical and genetic landscapes across 26 provinces. Our results support a hybrid model where demic diffusion predominantly impacts central China, while cultural diffusion and language assimilation occur in southwestern and coastal regions, respectively. This interdisciplinary study sheds light on the complex social-genetic history of the Han Chinese. By digitizing a large lexical dataset of Chinese dialects and comparing it to genetic profiles, Yang et al. reveal a hybrid model of language diffusion, consisting of both population migrations and social learning across different regions of China.
{"title":"Large-scale lexical and genetic alignment supports a hybrid model of Han Chinese demic and cultural diffusions","authors":"Chengkun Yang, Xiaoxi Zhang, Shi Yan, Sizhe Yang, Baihui Wu, Fengshuo You, Yue Cui, Ni Xie, Zhiyi Wang, Li Jin, Shuhua Xu, Menghan Zhang","doi":"10.1038/s41562-024-01886-9","DOIUrl":"10.1038/s41562-024-01886-9","url":null,"abstract":"The Han Chinese history is shaped by substantial demographic activities and sociocultural transmissions. However, it remains challenging to assess the contributions of demic and cultural diffusion to Han culture and language, primarily due to the lack of rigorous examination of genetic–linguistic congruence. Here we digitized a large-scale linguistic inventory comprising 1,018 lexical traits across 926 dialect varieties. Using phylogenetic analysis and admixture inference, we revealed a north–south gradient of lexical differences that probably resulted from historical migrations. Furthermore, we quantified extensive horizontal language transfers and pinpointed central China as a dialectal melting pot. Integrating genetic data from 30,408 Han Chinese individuals, we compared the lexical and genetic landscapes across 26 provinces. Our results support a hybrid model where demic diffusion predominantly impacts central China, while cultural diffusion and language assimilation occur in southwestern and coastal regions, respectively. This interdisciplinary study sheds light on the complex social-genetic history of the Han Chinese. By digitizing a large lexical dataset of Chinese dialects and comparing it to genetic profiles, Yang et al. reveal a hybrid model of language diffusion, consisting of both population migrations and social learning across different regions of China.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140914957","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01881-0
Anastasia Kozyreva, Philipp Lorenz-Spreen, Stefan M. Herzog, Ullrich K. H. Ecker, Stephan Lewandowsky, Ralph Hertwig, Ayesha Ali, Joe Bak-Coleman, Sarit Barzilai, Melisa Basol, Adam J. Berinsky, Cornelia Betsch, John Cook, Lisa K. Fazio, Michael Geers, Andrew M. Guess, Haifeng Huang, Horacio Larreguy, Rakoen Maertens, Folco Panizza, Gordon Pennycook, David G. Rand, Steve Rathje, Jason Reifler, Philipp Schmid, Mark Smith, Briony Swire-Thompson, Paula Szewach, Sander van der Linden, Sam Wineburg
The spread of misinformation through media and social networks threatens many aspects of society, including public health and the state of democracies. One approach to mitigating the effect of misinformation focuses on individual-level interventions, equipping policymakers and the public with essential tools to curb the spread and influence of falsehoods. Here we introduce a toolbox of individual-level interventions for reducing harm from online misinformation. Comprising an up-to-date account of interventions featured in 81 scientific papers from across the globe, the toolbox provides both a conceptual overview of nine main types of interventions, including their target, scope and examples, and a summary of the empirical evidence supporting the interventions, including the methods and experimental paradigms used to test them. The nine types of interventions covered are accuracy prompts, debunking and rebuttals, friction, inoculation, lateral reading and verification strategies, media-literacy tips, social norms, source-credibility labels, and warning and fact-checking labels. Kozyreva et al. review evidence from individual-level interventions for fighting online misinformation featured in 81 scientific papers. They classify the interventions in nine different types and summarize their findings in a toolbox.
{"title":"Toolbox of individual-level interventions against online misinformation","authors":"Anastasia Kozyreva, Philipp Lorenz-Spreen, Stefan M. Herzog, Ullrich K. H. Ecker, Stephan Lewandowsky, Ralph Hertwig, Ayesha Ali, Joe Bak-Coleman, Sarit Barzilai, Melisa Basol, Adam J. Berinsky, Cornelia Betsch, John Cook, Lisa K. Fazio, Michael Geers, Andrew M. Guess, Haifeng Huang, Horacio Larreguy, Rakoen Maertens, Folco Panizza, Gordon Pennycook, David G. Rand, Steve Rathje, Jason Reifler, Philipp Schmid, Mark Smith, Briony Swire-Thompson, Paula Szewach, Sander van der Linden, Sam Wineburg","doi":"10.1038/s41562-024-01881-0","DOIUrl":"10.1038/s41562-024-01881-0","url":null,"abstract":"The spread of misinformation through media and social networks threatens many aspects of society, including public health and the state of democracies. One approach to mitigating the effect of misinformation focuses on individual-level interventions, equipping policymakers and the public with essential tools to curb the spread and influence of falsehoods. Here we introduce a toolbox of individual-level interventions for reducing harm from online misinformation. Comprising an up-to-date account of interventions featured in 81 scientific papers from across the globe, the toolbox provides both a conceptual overview of nine main types of interventions, including their target, scope and examples, and a summary of the empirical evidence supporting the interventions, including the methods and experimental paradigms used to test them. The nine types of interventions covered are accuracy prompts, debunking and rebuttals, friction, inoculation, lateral reading and verification strategies, media-literacy tips, social norms, source-credibility labels, and warning and fact-checking labels. Kozyreva et al. review evidence from individual-level interventions for fighting online misinformation featured in 81 scientific papers. They classify the interventions in nine different types and summarize their findings in a toolbox.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140914861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01898-5
Yu Xu, Chuan-Chao Wang
The impetus behind the development of various Chinese dialects is as yet unknown. In a comprehensive quantitative coanalysis of linguistic and genetic data across China, Yang et al. find evidence to suggest that demographic diffusion, cultural diffusion and linguistic assimilation all contributed to the expansive diversity of Chinese dialects.
{"title":"Language evolution in China","authors":"Yu Xu, Chuan-Chao Wang","doi":"10.1038/s41562-024-01898-5","DOIUrl":"10.1038/s41562-024-01898-5","url":null,"abstract":"The impetus behind the development of various Chinese dialects is as yet unknown. In a comprehensive quantitative coanalysis of linguistic and genetic data across China, Yang et al. find evidence to suggest that demographic diffusion, cultural diffusion and linguistic assimilation all contributed to the expansive diversity of Chinese dialects.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140914884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01869-w
For patients affected by speech disorders, brain–machine-interface (BMI) devices could restore their ability to verbally communicate. In this work, we captured neural activity associated with internal speech — words said within the mind with no associated movement or audio output — and translated these cortical signals into text in real time.
{"title":"Brain–machine-interface device translates internal speech into text","authors":"","doi":"10.1038/s41562-024-01869-w","DOIUrl":"10.1038/s41562-024-01869-w","url":null,"abstract":"For patients affected by speech disorders, brain–machine-interface (BMI) devices could restore their ability to verbally communicate. In this work, we captured neural activity associated with internal speech — words said within the mind with no associated movement or audio output — and translated these cortical signals into text in real time.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140914980","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01867-y
Sarah K. Wandelt, David A. Bjånes, Kelsie Pejsa, Brian Lee, Charles Liu, Richard A. Andersen
Speech brain–machine interfaces (BMIs) translate brain signals into words or audio outputs, enabling communication for people having lost their speech abilities due to diseases or injury. While important advances in vocalized, attempted and mimed speech decoding have been achieved, results for internal speech decoding are sparse and have yet to achieve high functionality. Notably, it is still unclear from which brain areas internal speech can be decoded. Here two participants with tetraplegia with implanted microelectrode arrays located in the supramarginal gyrus (SMG) and primary somatosensory cortex (S1) performed internal and vocalized speech of six words and two pseudowords. In both participants, we found significant neural representation of internal and vocalized speech, at the single neuron and population level in the SMG. From recorded population activity in the SMG, the internally spoken and vocalized words were significantly decodable. In an offline analysis, we achieved average decoding accuracies of 55% and 24% for each participant, respectively (chance level 12.5%), and during an online internal speech BMI task, we averaged 79% and 23% accuracy, respectively. Evidence of shared neural representations between internal speech, word reading and vocalized speech processes was found in participant 1. SMG represented words as well as pseudowords, providing evidence for phonetic encoding. Furthermore, our decoder achieved high classification with multiple internal speech strategies (auditory imagination/visual imagination). Activity in S1 was modulated by vocalized but not internal speech in both participants, suggesting no articulator movements of the vocal tract occurred during internal speech production. This work represents a proof-of-concept for a high-performance internal speech BMI. Wandelt et al. describe a brain–machine interface that captures intracortical neural activity during internal speech (words said within the mind with no associated movement or audio output) and translates those cortical signals into real-time text.
{"title":"Representation of internal speech by single neurons in human supramarginal gyrus","authors":"Sarah K. Wandelt, David A. Bjånes, Kelsie Pejsa, Brian Lee, Charles Liu, Richard A. Andersen","doi":"10.1038/s41562-024-01867-y","DOIUrl":"10.1038/s41562-024-01867-y","url":null,"abstract":"Speech brain–machine interfaces (BMIs) translate brain signals into words or audio outputs, enabling communication for people having lost their speech abilities due to diseases or injury. While important advances in vocalized, attempted and mimed speech decoding have been achieved, results for internal speech decoding are sparse and have yet to achieve high functionality. Notably, it is still unclear from which brain areas internal speech can be decoded. Here two participants with tetraplegia with implanted microelectrode arrays located in the supramarginal gyrus (SMG) and primary somatosensory cortex (S1) performed internal and vocalized speech of six words and two pseudowords. In both participants, we found significant neural representation of internal and vocalized speech, at the single neuron and population level in the SMG. From recorded population activity in the SMG, the internally spoken and vocalized words were significantly decodable. In an offline analysis, we achieved average decoding accuracies of 55% and 24% for each participant, respectively (chance level 12.5%), and during an online internal speech BMI task, we averaged 79% and 23% accuracy, respectively. Evidence of shared neural representations between internal speech, word reading and vocalized speech processes was found in participant 1. SMG represented words as well as pseudowords, providing evidence for phonetic encoding. Furthermore, our decoder achieved high classification with multiple internal speech strategies (auditory imagination/visual imagination). Activity in S1 was modulated by vocalized but not internal speech in both participants, suggesting no articulator movements of the vocal tract occurred during internal speech production. This work represents a proof-of-concept for a high-performance internal speech BMI. Wandelt et al. describe a brain–machine interface that captures intracortical neural activity during internal speech (words said within the mind with no associated movement or audio output) and translates those cortical signals into real-time text.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.nature.com/articles/s41562-024-01867-y.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140914987","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-05-13DOI: 10.1038/s41562-024-01870-3
Qi Lin, Zifan Li, John Lafferty, Ilker Yildirim
Much of what we remember is not because of intentional selection, but simply a by-product of perceiving. This raises a foundational question about the architecture of the mind: how does perception interface with and influence memory? Here, inspired by a classic proposal relating perceptual processing to memory durability, the level-of-processing theory, we present a sparse coding model for compressing feature embeddings of images, and show that the reconstruction residuals from this model predict how well images are encoded into memory. In an open memorability dataset of scene images, we show that reconstruction error not only explains memory accuracy, but also response latencies during retrieval, subsuming, in the latter case, all of the variance explained by powerful vision-only models. We also confirm a prediction of this account with ‘model-driven psychophysics’. This work establishes reconstruction error as an important signal interfacing perception and memory, possibly through adaptive modulation of perceptual processing. Using a computational model to quantify difficulty in reconstructing images from compressed codes, Lin et al. show that reconstruction errors interface perception and memory by modulating how well images are encoded.
{"title":"Images with harder-to-reconstruct visual representations leave stronger memory traces","authors":"Qi Lin, Zifan Li, John Lafferty, Ilker Yildirim","doi":"10.1038/s41562-024-01870-3","DOIUrl":"10.1038/s41562-024-01870-3","url":null,"abstract":"Much of what we remember is not because of intentional selection, but simply a by-product of perceiving. This raises a foundational question about the architecture of the mind: how does perception interface with and influence memory? Here, inspired by a classic proposal relating perceptual processing to memory durability, the level-of-processing theory, we present a sparse coding model for compressing feature embeddings of images, and show that the reconstruction residuals from this model predict how well images are encoded into memory. In an open memorability dataset of scene images, we show that reconstruction error not only explains memory accuracy, but also response latencies during retrieval, subsuming, in the latter case, all of the variance explained by powerful vision-only models. We also confirm a prediction of this account with ‘model-driven psychophysics’. This work establishes reconstruction error as an important signal interfacing perception and memory, possibly through adaptive modulation of perceptual processing. Using a computational model to quantify difficulty in reconstructing images from compressed codes, Lin et al. show that reconstruction errors interface perception and memory by modulating how well images are encoded.","PeriodicalId":19074,"journal":{"name":"Nature Human Behaviour","volume":null,"pages":null},"PeriodicalIF":21.4,"publicationDate":"2024-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140914960","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}