Limits...
Mixed-complexity artificial grammar learning in humans and macaque monkeys: evaluating learning strategies.

Wilson B, Smith K, Petkov CI - Eur. J. Neurosci. (2015)

Bottom Line: We found no significant sensitivity to the non-adjacent AG relationships in the macaques.The results suggest that humans and macaques are largely comparably sensitive to the adjacent AG relationships and their statistical properties.However, in the presence of multiple cues to grammaticality, the non-adjacent relationships are less salient to the macaques and many of the humans.

View Article: PubMed Central - PubMed

Affiliation: Institute of Neuroscience, Newcastle University, Henry Wellcome Building, Framlington Place, Newcastle upon Tyne, NE2 4HH, UK; Centre for Behaviour and Evolution, Newcastle University, Newcastle upon Tyne, UK.

Show MeSH

Related in: MedlinePlus

Artificial grammar (AG) and stimulus sequences. (A) The AG contains five unique elements. Sequences (strings of nonsense words) consistent with the AG are generated by following any path of arrows from START to END. Violation sequences do not follow the arrows. The AG generates consistent sequences; however, all legal sequences must follow any of a number of ‘rules’, see text. The AG was used to generate eight exposure sequences, which follow all of these rules. Each experiment began with an exposure phase where the human or monkey participants passively listened to the habitation sequences for 5 or 30 min, respectively. (B) Four ‘consistent’ testing sequences (black) were generated from the AG. The first two of these sequences (in italics) were also presented in the exposure period (familiar), while the second two were novel to the testing phase of the experiments. Each consistent sequence was presented twice in each testing run, to balance the number of consistent and violation sequences presented. Eight violation sequences were generated, including different rule violations and a range of transitional probabilities (TPs; see Materials and methods). The sequences were designed in four pairs (denoted by colours). These pairs were matched as closely as possible for the rule violations they included and for their average TPs. Moreover, the second sequence in each pair, but not the first, violated the non‐adjacent, long‐distance, ‘ACF’ rule.
© Copyright Policy - creativeCommonsBy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4493314&req=5

ejn12834-fig-0001: Artificial grammar (AG) and stimulus sequences. (A) The AG contains five unique elements. Sequences (strings of nonsense words) consistent with the AG are generated by following any path of arrows from START to END. Violation sequences do not follow the arrows. The AG generates consistent sequences; however, all legal sequences must follow any of a number of ‘rules’, see text. The AG was used to generate eight exposure sequences, which follow all of these rules. Each experiment began with an exposure phase where the human or monkey participants passively listened to the habitation sequences for 5 or 30 min, respectively. (B) Four ‘consistent’ testing sequences (black) were generated from the AG. The first two of these sequences (in italics) were also presented in the exposure period (familiar), while the second two were novel to the testing phase of the experiments. Each consistent sequence was presented twice in each testing run, to balance the number of consistent and violation sequences presented. Eight violation sequences were generated, including different rule violations and a range of transitional probabilities (TPs; see Materials and methods). The sequences were designed in four pairs (denoted by colours). These pairs were matched as closely as possible for the rule violations they included and for their average TPs. Moreover, the second sequence in each pair, but not the first, violated the non‐adjacent, long‐distance, ‘ACF’ rule.

Mentions: The stimuli in the Rhesus macaque and the human experiments were identical. Each of the stimulus sequences (Fig. 1B) was created by digitally combining recordings of naturally spoken nonsense words produced by a female speaker based on an AG (Fig. 1A) developed by Saffran (2002) and (Saffran et al., 2008). Nonsense words were selected as stimuli because they have the advantage of being spectro‐temporally complex stimuli that are easy to distinguish from each other. The nonsense words were recorded with an Edirol R‐09HR (Roland) sound recorder. The amplitude of the recorded sounds was root‐mean‐square balanced. We computed the power spectra of the nonsense word stimuli and confirmed that they fall well within the auditory ranges of both species (Fig. S3). The nonsense word stimuli were randomly assigned to the AG elements (i.e. A = ‘yag’, C = ‘kem’, etc.). They were then combined into exposure and testing sequences using customised Matlab scripts (150 ms inter‐stimulus intervals). All the nonsense words were duration matched (413 ms), and each test sequence contained five nonsense word elements (sequence duration = 2665 ms).


Mixed-complexity artificial grammar learning in humans and macaque monkeys: evaluating learning strategies.

Wilson B, Smith K, Petkov CI - Eur. J. Neurosci. (2015)

Artificial grammar (AG) and stimulus sequences. (A) The AG contains five unique elements. Sequences (strings of nonsense words) consistent with the AG are generated by following any path of arrows from START to END. Violation sequences do not follow the arrows. The AG generates consistent sequences; however, all legal sequences must follow any of a number of ‘rules’, see text. The AG was used to generate eight exposure sequences, which follow all of these rules. Each experiment began with an exposure phase where the human or monkey participants passively listened to the habitation sequences for 5 or 30 min, respectively. (B) Four ‘consistent’ testing sequences (black) were generated from the AG. The first two of these sequences (in italics) were also presented in the exposure period (familiar), while the second two were novel to the testing phase of the experiments. Each consistent sequence was presented twice in each testing run, to balance the number of consistent and violation sequences presented. Eight violation sequences were generated, including different rule violations and a range of transitional probabilities (TPs; see Materials and methods). The sequences were designed in four pairs (denoted by colours). These pairs were matched as closely as possible for the rule violations they included and for their average TPs. Moreover, the second sequence in each pair, but not the first, violated the non‐adjacent, long‐distance, ‘ACF’ rule.
© Copyright Policy - creativeCommonsBy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4493314&req=5

ejn12834-fig-0001: Artificial grammar (AG) and stimulus sequences. (A) The AG contains five unique elements. Sequences (strings of nonsense words) consistent with the AG are generated by following any path of arrows from START to END. Violation sequences do not follow the arrows. The AG generates consistent sequences; however, all legal sequences must follow any of a number of ‘rules’, see text. The AG was used to generate eight exposure sequences, which follow all of these rules. Each experiment began with an exposure phase where the human or monkey participants passively listened to the habitation sequences for 5 or 30 min, respectively. (B) Four ‘consistent’ testing sequences (black) were generated from the AG. The first two of these sequences (in italics) were also presented in the exposure period (familiar), while the second two were novel to the testing phase of the experiments. Each consistent sequence was presented twice in each testing run, to balance the number of consistent and violation sequences presented. Eight violation sequences were generated, including different rule violations and a range of transitional probabilities (TPs; see Materials and methods). The sequences were designed in four pairs (denoted by colours). These pairs were matched as closely as possible for the rule violations they included and for their average TPs. Moreover, the second sequence in each pair, but not the first, violated the non‐adjacent, long‐distance, ‘ACF’ rule.
Mentions: The stimuli in the Rhesus macaque and the human experiments were identical. Each of the stimulus sequences (Fig. 1B) was created by digitally combining recordings of naturally spoken nonsense words produced by a female speaker based on an AG (Fig. 1A) developed by Saffran (2002) and (Saffran et al., 2008). Nonsense words were selected as stimuli because they have the advantage of being spectro‐temporally complex stimuli that are easy to distinguish from each other. The nonsense words were recorded with an Edirol R‐09HR (Roland) sound recorder. The amplitude of the recorded sounds was root‐mean‐square balanced. We computed the power spectra of the nonsense word stimuli and confirmed that they fall well within the auditory ranges of both species (Fig. S3). The nonsense word stimuli were randomly assigned to the AG elements (i.e. A = ‘yag’, C = ‘kem’, etc.). They were then combined into exposure and testing sequences using customised Matlab scripts (150 ms inter‐stimulus intervals). All the nonsense words were duration matched (413 ms), and each test sequence contained five nonsense word elements (sequence duration = 2665 ms).

Bottom Line: We found no significant sensitivity to the non-adjacent AG relationships in the macaques.The results suggest that humans and macaques are largely comparably sensitive to the adjacent AG relationships and their statistical properties.However, in the presence of multiple cues to grammaticality, the non-adjacent relationships are less salient to the macaques and many of the humans.

View Article: PubMed Central - PubMed

Affiliation: Institute of Neuroscience, Newcastle University, Henry Wellcome Building, Framlington Place, Newcastle upon Tyne, NE2 4HH, UK; Centre for Behaviour and Evolution, Newcastle University, Newcastle upon Tyne, UK.

Show MeSH
Related in: MedlinePlus