CoChat
Sign Up
Cited by 329Neural and Behavioral Psychology StudiesMemory Processes and InfluencesVisual Attention and Saliency Detection

Attention and associative learning in humans: An integrative review.

Open access
Le Pelley, Mike E., Mitchell, Chris J., Beesley, Tom, George, David N., Wills, Andy J.·2016·Psychological Bulletin
DOI 10.1037/bul0000064
Open this paper in CoChat — free accountOpen PDF

Abstract

. Rather, it suggests a new kind of “derived” attention.

AI snapshot

Written from the paper’s full text.

TL;DR

This integrative review argues that in humans, attention to cues is shaped by associative learning: predictive cues and cues that signal high-value outcomes receive more attention, through both controlled and automatic processes. It finds strong support for the predictiveness principle, weak support for the uncertainty principle, and proposes a simple model in which attention is determined by associative strength.

Key findings

  • There is strong evidence for the predictiveness principle: predictive cues are learned about faster and receive more overt attention, as shown by learned predictiveness, intradimensional–extradimensional shift, eye-tracking, attentional blink, and spatial cueing studies. In Denton and Kruschke (2006), test accuracy to the same low-salience green cue was 38% when it had been trained with a high-salience red cue versus 82% when trained with a low-salience red cue (chance = 33%).
  • Evidence for the uncertainty principle in humans is much weaker: neuroscientific, model-fitting, and qualitative findings are mixed, not always robust, and sometimes fail to replicate. Beesley et al. (2015) found greater overt attention to all cues under uncertain conditions (predictive cues predicted outcome with 67% accuracy) than certain conditions (100% accuracy), but the review concludes overall support for the uncertainty principle is not strong.
  • Learned value also influences attention: cues predicting high-value outcomes capture attention automatically, even when task-irrelevant and counterproductive. Le Pelley et al. (2015) found high-value distractors produced significantly more omission trials than low-value distractors; these effects persisted over training and were resistant to explicit instruction.
  • The authors propose a simple model in which attention to a cue is determined by its absolute associative strength, αA = |VA|, which accounts for most learned predictiveness and learned value data; the inverse base-rate effect is identified as an exception. Clinically, attentional bias predicts relapse, and Eberl et al. (2013) reported that cognitive retraining reduced one-year relapse rates by up to 10%.

Methods

  • Integrative/narrative review of human associative learning and attention, covering predictiveness, uncertainty, learned value, top-down/bottom-up attention, and a simple theoretical model.
  • Search 1: ISI Web of Science on 10/10/2015, in psychology, neuroscience, psychiatry, behavioral sciences, and multidisciplinary sciences, for articles citing Mackintosh (1975) or Pearce and Hall (1980) and containing keywords attention*, predict*, associab*, relevan*, irrelevan*, certain*, uncertain*; 828 hits, 28 additional articles found, yielding 143 relevant articles. Search 2: ISI Web of Science on 19/10/2015, for articles citing Anderson et al. (2011b), Della Libera and Chelazzi (2009), or Raymond and O’Brien (2009); 195 hits, 4 additional articles found, yielding 86 relevant articles.
  • Main analysis is a narrative synthesis of behavioral, eye-tracking, EEG/fMRI, model-fitting, and clinical studies, plus simulations of Mackintosh-style attention equations.

Limitations

  • Stated: evidence for the uncertainty principle in humans is not strong; model-fitting studies are mixed and do not test predictiveness-based alternatives; the inverse base-rate effect is an exception to the simple associative-strength model; future work is needed to determine when attentional exploration versus exploitation occurs.
  • Stated: many uncertainty-principle findings are not robust to replication, some unpublished replications are mixed, and Hogarth et al.’s findings conflict with those of Austin and Duka.
  • Evident: as a narrative review, it does not provide a systematic meta-analysis or effect-size synthesis; the search was limited to Web of Science and specific citing/keyword criteria, so relevant work outside those bounds may be missed.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1037/bul0000064

Cites

311 of 311 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 170 cited sources checked · 0 supported · 0 flagged

  • Findingcitation

    The accuracy of the response to exactly the same green stimulus depended on the salience of the stimulus that accompanied it during training.

    “Thus the accuracy of the response to exactly the same green stimulus depended on the salience of the stimulus that accompanied it during training.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Denton and Kruschke (2006)

      Not checked in the free run.

  • Findingcitation

    The ID-ED shift effect shows that learning can influence the attention paid to a stimulus dimension.

    “The ID-ED shift effect shows that learning can influence the attention that is paid to a stimulus dimension (e.g.colour, shape)”
    Quote in the paperNot a substantial claim
  • Interpretationreasoning

    Overall, current evidence suggests people tend to pay attention to cues that have reliably predicted valued outcomes in the past.

    “Overall, therefore, the current evidence suggests people tend to pay attention to cues that have reliably predicted valued outcomes in the past.”
    Quote in the paperSubstantial claim
  • Interpretationnone given

    People pay more attention to predictors of valuable outcomes than predictors of nonvaluable outcomes.

    “Here, we find that people pay more attention to predictors of valuable outcomes than predictors of nonvaluable outcomes.”
    Quote in the paperSubstantial claim
  • Interpretationreasoning

    The ID-ED shift effect is consistent with the idea of greater attention to predictive cues than nonpredictive cues.

    “The ID-ED shift effect is consistent with the idea of greater attention to predictive cues than nonpredictive cues.”
    Quote in the paperSubstantial claim
  • May combine two claims. 58
  • Classified as probably not a substantial claim. 28
  • The quote only partly states this claim. 2

  • May combine two claims.

    Findingnone given

    In the learned predictiveness effect, previously predictive cues are learned about more quickly than previously non-predictive cues, and they also command more attention as measured by tracking eye-gaze.

    “Here, previously predictive cues are learned about more quickly than previously non-predictive cues, and they also command more attention as measured by tracking eye-gaze.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning

    In that model, αA approaches the upper limit of 1 more rapidly when outcome magnitude is large (λ = 0.8) than when it is small (λ = 0.3).

    “it approaches this limit more rapidly when outcome magnitude is large”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning

    In the model where attention is determined by relative predictiveness, αA increases to the upper limit of 1 in both the large (λ = 0.8) and small (λ = 0.3) outcome magnitude cases.

    “A increases to the upper limit of 1 in both cases.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning

    In that model, asymptotic attention is greater when outcome magnitude is larger than when it is small.

    “asymptotic attention is greater when outcome magnitude is larger than when it is small.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In the simpler model, attention to a stimulus is determined by its absolute associative strength, which is in turn limited by the outcome magnitude λ.

  • May combine two claims.

    Interpretationreasoning

    In humans, there is a great deal of evidence in favour of the predictiveness principle and relatively little evidence to support the uncertainty principle.

    “We find that, in humans, there is a great deal of evidence in favour of the former, predictiveness principle and relatively little evidence to support the latter, uncertainty principle.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    There is evidence for both top-down and bottom-up attentional processes with respect to both predictiveness and learned value.

    “Another major finding from this survey is that there is evidence for both top-down and bottom-up attentional processes with respect to both predictiveness and learned value.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    People deliberately pay attention to cues that predict high-value outcomes, and attention is also captured by these cues outside of strategic control.

    “Hence, people deliberately pay attention to cues that predict high-value outcomes and, perhaps more interestingly, attention is also captured by these cues in a way that is outside of strategic control.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisereasoning

    The predictiveness principle holds that cues become more psychologically salient as a result of their predictiveness with respect to important outcomes, and more attention is allocated to predictive cues than to nonpredictive cues.

    “Consistent with this idea, the predictiveness principle refers to the idea that cues become more psychologically salient as a result of their predictiveness with respect to important outcomes; more attention will be allocated to predictive cues than to nonpredictive cues.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims. The quote only partly states this claim.

    Study design and conductown data

    The literature search on ISI Web of Science on 10/10/2015 yielded 828 hits, of which 143 articles were relevant to the current discussion.

    “The resulting 828 hits were then reviewed for content and for references to further literature”
    • Also stated in: § The Scope of the Current Survey · plain
      “yielding a total of 143 articles relevant to the current discussion”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The presence of the high-salience red pattern, but not the low-salience red pattern, overshadowed learning about the green pattern.

    “The implication is that the presence of the high-salience red pattern (but not the low-salience red pattern) overshadowed learning about the green pattern.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    The predictiveness principle and uncertainty principle are exemplified by theories proposed by Mackintosh (1975) and Pearce and Hall (1980) respectively.

    “These principles are exemplified by theories proposed by Mackintosh (1975) and Pearce and Hall (1980) respectively.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Mackintosh (1975)

      Not checked in the free run.

    • cites Pearce & Hall (1980)

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisenone given

    Past associative learning appears to change the way animals attend to their environment, showing that attention can be conditioned just as overt responses can be conditioned.

    “Here, past associative learning appears to change the way in which animals attend to their environment, showing that attention can be conditioned just as overt responses (lever-pressing, salivation etc) can be conditioned.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisenone given

    Two contrasting views of the relationship between predictiveness and attention have been dominant in the associative learning literature.

    “Two (contrasting) views of the relationship between predictiveness and attention have been dominant in the associative learning literature.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Stimulus selection can be driven by the perceptual properties of a stimulus; a stimulus is more likely to capture attention if it has an abrupt onset or bright colour.

    “It is well-known that stimulus selection can be driven by the perceptual properties of a stimulusa stimulus is more likely to capture attention if it has an abrupt onset (Yantis & Jonides, 1984), or bright colour (Theeuwes, 1992).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Yantis & Jonides, 1984

      Not checked in the free run.

    • cites Theeuwes, 1992

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    Directly analogous approaches have since been used to study the interaction between attention and learning in humans.

    “However, directly analogous approaches have since been used to study the interaction between attention and learning in humans.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Competition may be an important determinant of the extent to which learning takes place and of the way in which learning alters psychological salience by modulating attention.

    “Competition may be an important determinant, not only of the extent to which learning takes place, but also the way in which learning in turn alters psychological salience, by modulating attention.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning

    The learned predictiveness effect is not confounded with stimulus familiarity or differences in the objective cue-outcome relationships.

    “And like the ID-ED shift effect, the learned predictiveness effect is not confounded with stimulus familiarity (since all cues are presented the same number of times) or differences in the objective cue-outcome relationships (since all cues are equally predictive of the outcomes with which they are paired in Phase 2).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Stimulus intensity and physical salience can influence learning.

    “It is clear, then, that stimulus intensity and physical salience can influence learning.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    The allocation of attentional resources occurs on a continuum, such that differences in attention give rise to graded differences in the rate of learning about stimuli.

    “This later gave way to a less extreme view (see Sutherland & Mackintosh, 1971) on which the allocation of attentional resources occurs on a continuum, such that differences in attention give rise to graded differences in the rate of learning about stimuli, otherwise referred to as the associability or conditionability of those stimuli.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Sutherland & Mackintosh, 1971

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    The speed with which a cue-outcome association is learned can be taken as a measure of the attention paid to the cue.

    “Hence, the speed with which a cueoutcome association is learned can be taken as a measure of the attention paid to the cue.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Lashley (1929)

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    Such findings have since been replicated and extended in studies of human associative learning.

    “Such findings have since been replicated and extended in studies of human associative learning.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationnone given

    High intensity cues produce greater conditioned responding and prevent learning about other cues.

    “High intensity cues produce greater conditioned responding and prevent learning about other cues.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Participants in the Change group were aware of the predictive status of cues in Phase 1 and were able to use that knowledge to strategically allocate their attention to previously nonpredictive cues in Phase 2.

    “This finding shows that participants in the Change group were aware of the predictive status of cues in Phase 1 and were able to use that knowledge to strategically allocate their attention to previously nonpredictive cues in Phase 2.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The attentional capture by the predictive cue occurred even though it was not required by the task, was not adaptive with regard to that task, and the short SOA left little time for conscious processing.

    “This attentional capture occurred even though (1) it was not required by the task, (2) it was not adaptive with regard to that task, and (3) the short SOA meant that there was little time for participants to consciously process and respond to the stimuli on each test trial.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The bias observed in the attentional blink task was automatic and not under the participant's control.

    “Thus, there was no obvious advantage to the participant in biasing attention towards one stimulus and away from another; the implication is that the bias observed was automatic and not under the participant's control.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    Predictive cues are higher than non-predictive cues in both associability and overt attention.

    “This evidence, together with the earlier learned predictiveness findings, firmly establishes that predictive cues are higher than non-predictive cues in both associability and overt attention.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Subsequent studies using the same approach have found that, although instructions have a profound influence on the effect, there can also be a residual predictiveness effect that survives reversal instructions.

    “It should be noted, however, that subsequent studies using the same approach have found that, although instructions do have a profound influence on the effect, there can also be a residual predictiveness effect (a bias towards previously predictive cues when averaged across both groups) that survives reversal instructions”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Don & Livesey, 2015

      Not checked in the free run.

    • cites Shone et al., 2015

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Mitchell et al. (2012) saw no evidence for any automatic bias towards previously predictive cues; there was no evidence for any overall bias across both groups towards previously predictive cues.

    “Furthermore, Mitchell et al. (2012) saw no evidence for any automatic bias towards previously predictive cues; there was no evidence for any overall bias across both groups towards previously predictive cues.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Mitchell et al., 2012

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The predictive stimulus captured participants' spatial attention and hence sped responses to events occurring in that location.

    “It suggests that the predictive stimulus captured participants' spatial attention and hence sped responses to events occurring in that location; in this case, the onset of the probe.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    During Phase 2, participants in the Change group learned more about cues that had been nonpredictive in Phase 1, and paid greater overt attention to these cues, compared to those that had been predictive in Phase 1.

    “During Phase 2, participants in the Change group learned more about cues that had been nonpredictive in Phase 1, and paid greater overt attention to these cues (measured using eyetracking), compared to those that had been predictive in Phase 1.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Mitchell et al., 2012

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The attentional blink effect is widely assumed to reflect attentional limitations during encoding that restrict perceptual awareness of targets.

    “This effect is widely assumed to reflect attentional limitations during encoding that restrict perceptual awareness of targets”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Raymond, Shapiro & Arnell, 1992

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Studies of encoding of unsigned prediction errors in the brain are revealing in terms of functional brain anatomy, but they do not convincingly address the uncertainty principle.

    “In summary, while studies of encoding of uPEs in the brain are revealing in terms of functional brain anatomy, they do not convincingly address the uncertainty principle.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationcitation

    It is also possible that the greater variance explained by the more flexible model may reflect overfitting, fitting noise in addition to systematic patterns.

    “It is also possible that the greater variance explained by the more flexible model may reflect overfitting fitting noise in addition to systematic patterns”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Pitt & Myung, 2002

      Not checked in the free run.

    • cites Wills & Pothos, 2012

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    These studies fail to consider the relative adequacy of predictiveness- and uncertainty-based models, so the specific support offered by these model-fitting studies to the uncertainty principle remains rather weak.

    “In other words, these studies fail to consider the relative adequacy (Wills & Pothos, 2012) of predictiveness-and uncertainty-based models.Hence the specific support offered by these model-fitting studies to the uncertainty principle remains rather weak.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Wills & Pothos, 2012

      Not checked in the free run.

  • May combine two claims.

    Interpretationnone given

    The model-fitting studies cited above consider only two options, a fixed-rate model versus a variable-rate model incorporating the uncertainty principle, and do not consider alternative versions of a variable-rate model, in particular a version incorporating the predictiveness principle.

    “Perhaps more importantly, the model-fitting studies cited above consider only two options: a fixed-rate model versus a variable-rate model incorporating the uncertainty principle.They do not consider alternative versions of a variable-rate model; in particular, a version incorporating the predictiveness principleperhaps such a model would provide the best fit of all.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    The critical effect of surprising trials on later learning is typically weak, and not always replicated.

    “In contrast, the critical effect of the surprising trials on later learning is typically weak, and not always replicated.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Unpublished attempts to replicate the negative transfer effect of Griffiths et al. (2011), by the original researchers, have been mixed.

  • May combine two claims.

    Premisecitation

    In rats, there is evidence in support of both the uncertainty principle and the predictiveness principle.

    “In rats, there is evidence in support of both the uncertainty principle and the predictiveness principle”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Le Pelley, 2004

      Not checked in the free run.

    • cites Pearce & Mackintosh, 2010

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Findings are mixed and not all possible models, such as predictiveness-based models, have been tested.

    “However, findings are mixed and not all possible models (e.g., predictiveness-based models) have been tested.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    These latter imaging data show a link between cues and unsigned prediction errors, but they still fall short of confirming the uncertainty principle.

    “While these latter imaging data show a link between cues and uPEs, they still fall short of confirming the uncertainty principle.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Several fMRI studies with humans have found brain regions whose activity correlates across trials with unsigned prediction error, with greater activity generated by presentation of surprising outcomes than predicted outcomes.

    “Several fMRI studies with humans have found brain regions whose activity correlates across trials with uPE, with greater activity generated by presentation of surprising outcomes than predicted outcomes.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Boll, Gamer, Gluth, Finsterbusch & Buchel, 2013

      Not checked in the free run.

    • cites Prevost, McCabe, Jessup, Bossaerts & O'Doherty, 2011

      Not checked in the free run.

    • cites Ploghaus et al., 2000

      Not checked in the free run.

  • May combine two claims.

    Premisereasoning

    The fixed-rate model is a special case of a variable-rate model in which the parameter controlling changes in rate is set to zero, so the variable-rate model will always fit at least as well as the fixed-rate model.

    “The fixed-rate model is a special case of a variable-rate model in which the parameter controlling changes in rate is set to zero, so the variable-rate model will always fit at least as well as the fixed-rate model.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Whether a fixed- or variable-rate model is preferred comes down to how heavily the greater flexibility of the variable model is penalized, and there is debate over the most appropriate way to implement this penalty.

    “So in the end, whether a fixed-or variable-rate model is preferred comes down to how heavily the greater flexibility of the variable model is penalized; there is debate over the most appropriate way to implement this penalty”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Kenny, 2015

      Not checked in the free run.

    • cites Pitt & Myung, 2002

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisenone given

    Some studies have found that models in which attention to cues is variable, and determined by uncertainty, fit empirical learning data better than models in which attention is fixed.

    “Some studies have found that models in which attention to cues is variable, and determined by uncertainty, fit empirical learning data better than models in which attention is fixed.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    Studies cited above show that unsigned prediction error is encoded in the brain, which is a prerequisite of the uncertainty principle.

    “Such studies show that uPE is encoded in the brain, which is a prerequisite of the uncertainty principle.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationreasoning

    The reason for the difference in highlighted brain regions across studies of unsigned prediction error is not clear, though it most likely relates to the wide variety of different procedures.

    “The reason for the difference in highlighted brain regions across these studies is not clear, though it most likely relates to the wide variety of different procedures.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Overt attention was greater to less predictive cues than to more predictive cues.

    “The implication is that overt attention was greater to less predictive cues than to more predictive cues”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    This finding of greater attention to cues under conditions of greater uncertainty accords with the uncertainty principle embodied by the Pearce-Hall model and with the suggestion that uncertainty drives attentional exploration.

    “This finding of greater attention to cues under conditions of greater uncertainty accords with the uncertainty principle embodied by the Pearce-Hall model, and with the suggestion that uncertainty drives attentional exploration of cues for prediction-relevant information.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    All studies supporting the uncertainty principle have used procedures in which the outcome cannot be predicted perfectly on each trial.

    “As far as we are aware, it is true that all studies supporting the uncertainty principle have used procedures in which the outcome cannot be predicted perfectly on each trial.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    The relative lack of evidence for the Pearce-Hall process may simply reflect a trend in the types of procedures that have been used to study attentional learning in humans.

    “However, the relative lack of evidence for the Pearce-Hall process may simply reflect a trend in the types of procedures that have been used to study attentional learning in humans.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: In studies that do not involve substantial monetary rewards, participants' primary motivation is to complete the experiment as quickly as possible and with minimum effort.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    There are certain phenomena demonstrating a relationship between learning and attention that are anticipated by both the predictiveness principle and the uncertainty principle.

    “Before moving on, we note briefly that there are certain phenomena demonstrating a relationship between learning and attention that are anticipated by both the predictiveness principle and the uncertainty principle.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    It also follows from the uncertainty principle because, on early extinction trials, the occurrence of 'no illness' after eating steak at restaurant R is surprising and hence will result in an increase in attention to all presented stimuli, which includes the restaurant.

    “It also follows from the uncertainty principle because, on early extinction trials, the occurrence of 'no illness' after eating steak at restaurant R is surprising and hence (according to the Pearce-Hall model) will result in an increase in attention to all presented stimuli, which includes the restaurant.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    This finding suggests that learning about the rewards associated with stimuli modulated attention to those stimuli, with greater attention allocated to the high-reward target colour than the low-reward colour.

    “This finding suggests that learning about the rewards associated with stimuli modulated attention to those stimuli; greater attention allocated to the high-reward target colour than the low-reward colour.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Attentional capture by the high-value colour during the test phase may not be driven by its learned value, but because this colour defined the target during the preceding training phase.

    “Attentional capture by the highvalue colour (say, red) during the test phase may not be driven by its learned value, but because this colour defined the target during the preceding training phase.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    Hybrid models incorporating both the predictiveness and uncertainty principles have been put forward in the context of animal learning theory.

    “Hybrid models incorporating both of these principles have been put forward in the context of animal learning theory, though it remains for future empirical work to establish more convincingly whether both of these mechanisms operate in humans.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationnone given

    While attention tends to increase towards cues that reliably signal their outcomes, even greater attention will be paid to cues that reliably signal the outcomes that have the highest value.

    “That is, while attention tends to increase towards cues that reliably signal their outcomes, even greater attention will be paid to cues that reliably signal the outcomes that have the highest value.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    This suggestion is consistent with the original characterizations of uncertainty-driven and predictiveness-driven processes in the models of Pearce & Hall (1980) and Mackintosh (1975) and offers one way the predictiveness and uncertainty principles might be reconciled.

    “This suggestion is consistent with the original characterizations of uncertainty-driven and predictiveness-driven processes in the models of Pearce & Hall (1980) and Mackintosh (1975) respectively, and offers one way in which the predictiveness and uncertainty principles-which at first glance seem to be mutually contradictory-might be reconciled (see Le Pelley, 2004, 2010, for a hybrid model along”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Pearce & Hall (1980)

      Not checked in the free run.

    • cites Mackintosh (1975)

      Not checked in the free run.

    • cites Le Pelley, 2004, 2010

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Interpretationcitation

    Mackintosh's (1975) model anticipates greater attention to stimuli that predict a high-value outcome than to those that predict a low-value outcome early in training, but at asymptote attention depends on learned predictiveness and not learned value.

    “Consequently, at least early in training this model correctly anticipates greater attention to stimuli that predict a high-value outcome (e.g., a large reward) than those that predict a low-value outcome (Figure 3a).However, at asymptote the model anticipates that attention will depend on learned predictiveness (attention will be greater to predictive than nonpredictive cues) but not learned value”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Mackintosh, 1975

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Learned signal value can modulate the strength of capture caused by physical salience and can drive capture by non-physically salient stimuli in its own right.

    “This last finding suggests that learned signal value can modulate the strength of capture that would otherwise by caused by physical salience, and drive capture by non-physically salient stimuli in its own right.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    In procedures where reward-predictive stimuli are task-irrelevant throughout, attentional orienting to high-value cues is never rewarded, so an attentional habit cannot have developed; the bias must be a consequence of the signal value of the cues.

    “Here, attentional orienting to the high-value cues is never rewarded and so it is not possible for an attentional habit to have developed.”
    • Also stated in: § Summary: Learned Value Affects Top-Down and Bottom-Up Attention · plain
      “The attentional bias must, therefore, be a consequence of the signal value of the cues and hence their ability to capture attention.”
    Quote in the paperSubstantial claim

  • May combine two claims. The quote only partly states this claim.

    Premisecitation

    In Le Pelley et al. (2015), high-value distractors produced significantly more omission trials than low-value distractors; people were more likely to look at high-value distractors even though doing so lost them a large reward.

    “That is, people were more likely to look at high-value distractors than low-value distractors, even though doing so lost them the large reward that they would otherwise have obtained.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Le Pelley et al., 2015

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    It is not possible to say for certain whether VMAC is due to a habitual orienting response or to a change in effective salience of stimuli, because both can explain the observed data.

    “It is not possible to say for certain which of these two potential mechanisms is responsible for the effects seen in the training phasetest phase and the intertrial priming procedures described aboveboth can explain the observed data.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Neither Anderson et al. nor Qi et al. tested whether capture by the former target was significantly greater in a rewarded procedure than in an unrewarded procedure, so these studies do not provide a robust test of the claim that reward is the critical variable.

    “Neither Anderson et al. nor Qi et al. tested whether capture by the former target was significantly greater in a rewarded procedure than in an unrewarded procedure, so these studies do not provide a robust test of the claim that reward is the critical variable.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    In inter-trial priming, target choice on the previous trial influences performance on the current trial, producing a benefit if the same target repeats and a decrement if a different target appears with the previous target as distractor.

    “Suppose that participants must respond to a red target on trial T; this will produce a performance benefit on the next trial (T+1) if it also features a red target, but a performance decrement if trial T+1 has a green target while a red stimulus appears as a distractor (Maljkovic & Nakayama, 1994, 1996, 2000).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Maljkovic & Nakayama, 1994, 1996, 2000

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Learning of stable reward associations can exert a long-term influence on attentional capture, while experience of individual and unpredictable stimulus-reward pairings can have a similar but short-term effect on capture.

    “In combination, then, it seems that learning of 'stable' reward associations can exert a long-term influence on attentional capture, while experience of individual and unpredictable stimulus-reward pairings can have a similar, but short-term, effect on capture.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Following an action with reward increases the likelihood that the action will occur in the future (the Law of Effect).

    “A fundamental principal of reward learning is that following an action with reward increases the likelihood that the action will occur in the future (the so-called Law of Effect : Thorndike, 1911).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Thorndike, 1911

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The implication is that capture by the former high-reward target in the rewarded version of the task is a product of its learned value, and not merely because it has previously been a target.

    “The implication is that capture by the former high-reward target in the rewarded version of the task is a product of its learned value, and not merely because it has previously been a target.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Wang et al. tested the critical interaction between rewarded and unrewarded procedures and found it to be significant.

    “Wang et al. are alone in testing this critical interaction, which they did find to be significant.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Wang et al.

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Stimuli with high learned value captured attention and reduced attentional resources available for processing the target; this capture effect was nonspatial because all stimuli were presented in the same location.

    “As for the spatial studies discussed earlier, this suggests that stimuli with high learned value captured attention and so reduced attentional resources available for processing the target.However here this capture effect must have been nonspatial, since all stimuli were presented in the same location.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Using EEG, Wills et al. (2014) showed that a validated neural marker of selective attention was modulated by IBRE training, consistent with greater attention to C than to B.

    “Furthermore, using EEG, Wills et al. (2014) showed that a validated neural marker of selective attention was also modulated by IBRE training, consistent with greater attention to C than to B.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Wills et al., 2014

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationcitation

    In Failing and Theeuwes (2015), it is possible that greater selection history, rather than difference in learned value, drives greater capture by high-value stimuli on test.

    “raising the possibility that it is this greater selection history (rather than the difference in learned value) that drives greater capture by these stimuli on test.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Failing and Theeuwes (2015)

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    There is abundant evidence for attentional learning in humans; attention is influenced by learning about the predictiveness of a stimulus and about its value, and both phenomena manifest as top-down and bottom-up attentional biases.

    “There is an abundance of evidence for attentional learning in humans.Attention is influenced both by learning about the predictiveness of a stimulus (how much information it provides about other, outcome events) and about its value (how significant-rewarding, punishingthose outcome events are).Furthermore, both of these phenomena manifest themselves as both top-down (controlled) and bottom-up (aut”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationnone given

    Most of the data presented in the review are captured by the idea that learning increases attention to cues to the extent that those cues acquire associative strength; the IBRE represents the single exception of which the authors are aware.

    “Most of the data presented in this review are captured by the idea that learning increases attention to cues to the extent that those cues acquire associative strength.The IBRE represents the single exception to this rule of which we are aware.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    The IBRE can be observed even when C has, if anything, a lower associative strength than B; therefore the attentional advantage for C over B cannot easily be explained by a model equating attention with associative strength.

    “So the IBRE can be observed even when C has (if anything) a lower associative strength than B. As such, we cannot easily explain an attentional advantage for C over B in terms of our model which equates attention with associative strength.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The model in which attention is determined by relative predictiveness anticipates an influence of learned value on attention early in training, but not at asymptote.

    “So this model anticipates an influence of learned value on attention early in training, but not at asymptote.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    According to the uncertainty principle, surprise should increase uncertainty regarding the predictiveness of the target cue, which should produce an increase in attention and hence increase the speed with which the new relationship between the target and the strong outcome is learned in Phase 2.

    “According to the uncertainty principle this should produce an increase in attention, and hence increase the speed with which the new relationship between the target and the strong outcome is learned in Phase 2.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Automatic attention to cues that predict valued outcomes is not goal-directed or under top-down control, nor is it driven by the intrinsic physical properties of the stimulus; the change in effective salience of the cue results from training and experience.

    “The automatic attention that participants pay to cues that predict valued outcomes is not goal-directed or under top-down control, but neither is it driven by the intrinsic, physical properties of the stimulus.Rather, the change in effective salience of the cue is the result of training; it comes about as a consequence of the participant's experience with, and memory of, that cue and its”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    The concept of derived attention does not fit the standard goal-directed/stimulus-driven distinction.

    “does not fit the standard goal-directed/stimulus-driven distinction.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    In many cases where an IBRE is observed, accuracy in predicting the common outcome for cue B alone numerically exceeds accuracy for predicting the rare outcome for cue C alone; in one case this advantage for B over C was statistically significant.

    “In many (but not all) cases where an IBRE is observed, accuracy in predicting the common outcome for cue B alone numerically exceeds accuracy for predicting the rare outcome for cue C alone (Bohil, Markman & Maddox, 2005;Medin & Edelson, 1988;Wills et al., 2014;Winman, Wennerholm, Juslin & Shanks, 2005); in one case (Wills et al., 2014) this advantage for B over C was tested statistically, and”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Bohil, Markman & Maddox, 2005

      Not checked in the free run.

    • cites Medin & Edelson, 1988

      Not checked in the free run.

    • cites Wills et al., 2014

      Not checked in the free run.

  • May combine two claims.

    Premisereasoning

    In some studies, the outcome valence (gain or loss of money) and the probability with which that outcome occurred were confounded, so it is unclear which was responsible for the differences in attention observed during the test phase.

    “outcome they were paired with (gain or loss of money), and the probability with which that outcome occurred.Hence it is unclear which was responsible for the differences in attention observed during the test phase of these studies.”
    Quote in the paperSubstantial claim
79 of 79

  • The paper relies on this without support.

    Research concerning interactions between attention and learning in humans has proceeded largely independently in two different psychological traditions.

    “Interestingly, research concerning interactions between attention and learning in humans has (to date) proceeded largely independently in two different psychological traditions.”

  • The authors state this as an assumption.

    The instructions given to the Change and Continuity groups were equally effective in producing top-down changes in attention.

    “However, the validity of this conclusion rests on the assumption that the instructions given to the Change and Continuity groups were equally effective in producing top-down changes in attention, but this may not be the case.”

  • The paper relies on this without support.

    Unpublished attempts to replicate the negative transfer effect of Griffiths et al. (2011), by the original researchers, have been mixed.

    “It should be noted, however, that unpublished attempts to replicate the negative transfer effect of Griffiths et al. (2011), by the original researchers, have been mixed.”

  • The paper relies on this without support.

    Training with a moderate outcome slows later learning with a larger outcome.

    “There is no doubt that training with a moderate outcome slows later learning with a larger outcome.”

  • The authors state this as an assumption.

    In studies that do not involve substantial monetary rewards, participants' primary motivation is to complete the experiment as quickly as possible and with minimum effort.

    “In studies that do not involve substantial monetary rewards (the vast majority of studies of predictiveness in humans), it seems fair to assume that participants' primary motivation is to complete the experiment as quickly as possible and with minimum effort.”

  • The authors state this as an assumption.

    Singleton search primarily reflects the operation of rapid, bottom-up attentional processes due to the physical salience of the target.

    “It is often assumed that such singleton search primarily reflects the operation of rapid, bottom-up attentional processes due to the physical salience of the target, so it seems plausible that reward might be modulating the operation of this bottomup process (see also O'Brien & Raymond, 2012;Stankevich & Geng, 2015, for further evidence consistent with a rapid influence of learned value).”

  • The authors state this as an assumption.

    In the simpler model, attention to a stimulus is determined by its absolute associative strength, which is in turn limited by the outcome magnitude λ.

    “since attention is determined by associative strength, which is in turn limited by”
7 of 7

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Mackintosh's attention rule is consistent with a number of results, and contemporary approaches to attention and learning suppose that both the Mackintosh and Pearce-Hall rules can play a role in any situation.

    “Mackintosh’s attention rule is consistent with a number of results (e.g., Le Pelley et al., 2016; Pearce & Mackintosh, 2010), and contemporary approaches to attention and learning now suppose that both the Mackintosh and Pearce-Hall rules can play a role in any situation (e.g., Le Pelley, 2004; Pearce & Mackintosh, 2010; see Chao et al., 2021, for recent discussion).”

    From Context, attention, and the switch between habit and goal-direction in behavior (2021) · cites this paper as Le Pelley et al., 2016

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Phase 1 example uses a Phase 2 reaction

    Minor

    The paper says Mr X's Phase 1 reactions were sweating or dizziness, but later describes nausea as a Phase 1 reaction. Nausea is introduced as a reaction for Mr Y in Phase 2.

  • Four additional articles are missing from the Search 2 tally

    Minor

    The exclusion categories total 109 articles, leaving 86 of the original 195 hits relevant. The paper also says it found four other relevant articles through references, but still reports 86 relevant articles in total; including those four would make 90.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 12

  1. Shone, Harris & Livesey, 2015: see also Kersten, Goldstone & Schaffert, 1998) 3 Some studies of blocking have found evidence of greater overt attention to the pretrained cue A than to the blocked cue B (Ellis et al., 2014; Le Pelley et al., 2007). This finding is consistent with the predictiveness principle (since pretraining renders A a better predictor of the outcome), but not with the uncertainty principle (since pretraining means that the outcome on AB trials is, if anything, less surprising given the presence of A than the presence of B). However, the difference in predictiveness between A and B is confounded with a difference ©2016 American Psychological Association. This article may not exactly replicate the authoritative document published in the APA journal. It is not the copy of record. in familiarity (pretraining means that A is experienced more times than has B). Given the possibility that familiarity might exert an independent effect on attention, the finding of greater attention to A than B provides somewhat ambiguous support for the predictiveness principle over the uncertainty principle. 4 (Balcetis, Dunning & Granot, 2012; Brielmann & Spering, 2015;
  2. Chapman, Gallivan, Wong, et al., 2015; O'Brien & Raymond, 2012; Raymond & O'Brien, 2009; Rothkirch et al., 2013; see also Rutherford, O'Brien & Raymond, 2010) conflated learned value and learned predictiveness: stimuli differed both in the nature of the ©2016 American Psychological Association. This article may not exactly replicate the authoritative document published in the APA journal. It is not the copy of record. References a marks articles retrieved from the ISI Web of Science database search for articles on learned predictiveness (Search 1 -see main text); b marks additional articles on learned predictiveness found via references in articles from this search. c marks articles retrieved from the ISI Web of Science database search for articles on learned value (Search 2 -see main text); d marks additional articles on learned value found via references in articles from this search.
  3. Chun, M.M., & Potter, M.C. (1995). A two-stage model for multiple target detection in RSVP. Journal of Experimental Psychology: Human Perception and Performance, 21, 109-127.
  4. Holland, P.C., & Maddux, J.-M. (2010). Brain systems of attention in associative learning. In C.J. Mitchell & M.E. Le Pelley (Eds.), Attention and Associative Learning: From Brain to Behaviour (pp. 305-349). Oxford, UK: Oxford University Press.
  5. Holmes, N.M., & Harris, J.A. (2010). Latent inhibition. In C.J. Mitchell & M.E. Le Pelley (Eds.), Attention and associative learning: From brain to behaviour (pp. 99-130). Oxford, UK: Oxford University Press.
  6. Kruschke, J.K. (2009). Highlighting: A canonical experiment. In B. Ross (Ed.), The Psychology of Learning and Motivation (Vol. 51, pp. 153-185). San Diego, CA: Academic Press.
  7. Marissen, M.A.E., Franken, I.H.A., Waters, A.J., Blanken, P., van den Brink, W., & Hendriks, ©2016 American Psychological Association. This article may not exactly replicate the authoritative document published in the APA journal. It is not the copy of record.
  8. Pavlov, I.P. (1927). Conditioned reflexes. London: Oxford University Press.
  9. Pearce, J.M., George, D.N., & Redhead, E.S. (1998). The role of attention in the solution of ©2016 American Psychological Association. This article may not exactly replicate the authoritative document published in the APA journal. It is not the copy of record. conditional discriminations. In N.A. Schmajuk & P.C. Holland (Eds.), Occasion setting: Associative learning and cognition in animals. Washington, D.C.: American Psychological Association.
  10. Posner, M.I., Nissen, M.J., & Ogden, W.C. (1978). Attended and unattended processing modes: The role of set for spatial location. In H.L. Pick & I.J. Saltzman (Eds.), Modes of perceiving and processing information. (pp. 137-157). Hillsdale, NJ: Erlbaum.
  11. Siddle, D.A.T., Booth, M.L., & Packer, J.S. (1987). Effects of stimulus preexposure on omission responding and omission-produced dishabituation of the human electrodermal response. Quarterly Journal of Experimental Psychology Section B-Comparative and ©2016 American Psychological Association. This article may not exactly replicate the authoritative document published in the APA journal. It is not the copy of record. Physiological Psychology, 39, 339-363.
  12. As A develops associative strength, A increases for both  = 0.8 and  = 0.3. However, since attention is determined by associative strength, which is in turn limited by , asymptotic attention is greater when outcome magnitude is larger than when it is small. So this simpler model anticipates a persistent influence of learned value on attention. [Other parameter: S = 0.3. (Figure taken from Le Pelley, Beesley & Griffiths, in press) ©2016 American Psychological Association. This article may not exactly replicate the authoritative document published in the APA journal. It is not the copy of record.

Cited by

Sorted by citations

Related

Authors
Le Pelley, M. E., Mitchell, C. J., Beesley, T., George, D. N., & Wills, A. J.
Affiliations
  • University of Plymouth (GB)
  • UNSW Sydney (AU)
  • University of Hull (GB)
ORCID
Venue
Psychological Bulletin
Publisher
American Psychological Association (APA)
Published
2016-10
Volume
142
Issue
10
Pages
1111-1140
ISSN
  • 0033-2909 (ISSN-L)
  • 1939-1455
DOI
10.1037/bul0000064
OpenAlex
W2494089081
Semantic Scholar
5345fafe1c561edf9a874d44ae30da5ac1e797cf
Cited by
329
References
297
Citations by year
2026: 21 · 2025: 29 · 2024: 33 · 2023: 23 · 2022: 29 · 2021: 49 · 2020: 37 · 2019: 38 · 2018: 32 · 2017: 32 · 2016: 5
FWCI
22.70
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Psychological Bulletin (publishedVersion)
  • PubMed (publishedVersion)
  • Repository@Hull (Worktribe) (University of Hull) (submittedVersion, cc-by)
  • Repository@Hull (Worktribe) (University of Hull) (submittedVersion, other-oa)
  • PEARL (University of Plymouth) (submittedVersion)
  • Lancaster EPrints (Lancaster University) (submittedVersion)
Funders
University of Cambridge
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural and Behavioral Psychology Studies
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Memory Processes and Influences
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Visual Attention and Saliency Detection
Keywords
attention, associative learning, attentional control, cognitive control
MeSH terms
Association Learning, Attention, Humans

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review