Limiting factors for mapping corpus-based semantic representations to brain activity

Research output: Contribution to journalArticlepeer-review

Standard

Limiting factors for mapping corpus-based semantic representations to brain activity. / Bullinaria, J.A.; Levy, J.P.

In: PLoS ONE, Vol. 8, No. 3, e57191, 19.03.2013.

Research output: Contribution to journalArticlepeer-review

Harvard

APA

Vancouver

Author

Bibtex

@article{ee314ea38c79402ab223ed3c6a7aae48,
title = "Limiting factors for mapping corpus-based semantic representations to brain activity",
abstract = "To help understand how semantic information is represented in the human brain, a number of previous studies have explored how a linear mapping from corpus derived semantic representations to corresponding patterns of fMRI brain activations can be learned. They have demonstrated that such a mapping for concrete nouns is able to predict brain activations with accuracy levels significantly above chance, but the more recent elaborations have achieved relatively little performance improvement over the original study. In fact, the absolute accuracies of all these models are still currently rather limited, and it is not clear which aspects of the approach need improving in order to achieve performance levels that might lead to better accounts of human capabilities. This paper presents a systematic series of computational experiments designed to identify the limiting factors of the approach. Two distinct series of artificial brain activation vectors with varying levels of noise are introduced to characterize how the brain activation data restricts performance, and improved corpus based semantic vectors are developed to determine how the word set and model inputs affect the results. These experiments lead to the conclusion that the current state-of-the-art input semantic representations are already operating nearly perfectly (at least for non-ambiguous concrete nouns), and that it is primarily the quality of the fMRI data that is limiting what can be achieved with this approach. The results allow the study to end with empirically informed suggestions about the best directions for future research in this area.",
keywords = "Conceptual semantics, Forecasting, Functional magnetic resonance imaging, Human learning, Human performance, Learning, Lexical semantics, Semantics",
author = "J.A. Bullinaria and J.P. Levy",
year = "2013",
month = mar,
day = "19",
doi = "10.1371/journal.pone.0057191",
language = "English",
volume = "8",
journal = "PLoSONE",
issn = "1932-6203",
publisher = "Public Library of Science (PLOS)",
number = "3",

}

RIS

TY - JOUR

T1 - Limiting factors for mapping corpus-based semantic representations to brain activity

AU - Bullinaria, J.A.

AU - Levy, J.P.

PY - 2013/3/19

Y1 - 2013/3/19

N2 - To help understand how semantic information is represented in the human brain, a number of previous studies have explored how a linear mapping from corpus derived semantic representations to corresponding patterns of fMRI brain activations can be learned. They have demonstrated that such a mapping for concrete nouns is able to predict brain activations with accuracy levels significantly above chance, but the more recent elaborations have achieved relatively little performance improvement over the original study. In fact, the absolute accuracies of all these models are still currently rather limited, and it is not clear which aspects of the approach need improving in order to achieve performance levels that might lead to better accounts of human capabilities. This paper presents a systematic series of computational experiments designed to identify the limiting factors of the approach. Two distinct series of artificial brain activation vectors with varying levels of noise are introduced to characterize how the brain activation data restricts performance, and improved corpus based semantic vectors are developed to determine how the word set and model inputs affect the results. These experiments lead to the conclusion that the current state-of-the-art input semantic representations are already operating nearly perfectly (at least for non-ambiguous concrete nouns), and that it is primarily the quality of the fMRI data that is limiting what can be achieved with this approach. The results allow the study to end with empirically informed suggestions about the best directions for future research in this area.

AB - To help understand how semantic information is represented in the human brain, a number of previous studies have explored how a linear mapping from corpus derived semantic representations to corresponding patterns of fMRI brain activations can be learned. They have demonstrated that such a mapping for concrete nouns is able to predict brain activations with accuracy levels significantly above chance, but the more recent elaborations have achieved relatively little performance improvement over the original study. In fact, the absolute accuracies of all these models are still currently rather limited, and it is not clear which aspects of the approach need improving in order to achieve performance levels that might lead to better accounts of human capabilities. This paper presents a systematic series of computational experiments designed to identify the limiting factors of the approach. Two distinct series of artificial brain activation vectors with varying levels of noise are introduced to characterize how the brain activation data restricts performance, and improved corpus based semantic vectors are developed to determine how the word set and model inputs affect the results. These experiments lead to the conclusion that the current state-of-the-art input semantic representations are already operating nearly perfectly (at least for non-ambiguous concrete nouns), and that it is primarily the quality of the fMRI data that is limiting what can be achieved with this approach. The results allow the study to end with empirically informed suggestions about the best directions for future research in this area.

KW - Conceptual semantics

KW - Forecasting

KW - Functional magnetic resonance imaging

KW - Human learning

KW - Human performance

KW - Learning

KW - Lexical semantics

KW - Semantics

UR - http://www.scopus.com/inward/record.url?partnerID=yv4JPVwI&eid=2-s2.0-84875097974&md5=9f0fd6da2ceb5352293d3cc04b367ec4

U2 - 10.1371/journal.pone.0057191

DO - 10.1371/journal.pone.0057191

M3 - Article

C2 - 23526937

AN - SCOPUS:84875097974

VL - 8

JO - PLoSONE

JF - PLoSONE

SN - 1932-6203

IS - 3

M1 - e57191

ER -