An Introduction to ERP Studies of Attention Logan Trujillo, Ph.D. Post-Doctoral Fellow University of Texas at Austin Cognitive Science Course, Fall 2008
What is Attention? Everyone knows what attention is. It is the taking possession by the mind, in clear and vivid form, of one out of what seem several simultaneously possible objects or trains of thought. Focalization, concentration, of consciousness are of its essence. It implies withdrawal from some things in order to deal effectively with others William James (1890)
How is Attention studied in Psychology & Cognitive Neuroscience? Behavioral paradigms useful, but some limitations EX: locus of selection does attention influence sensory processing or late stage cognition for a given task, or both? Difficult to tell from RT & accuracy data alone. Brain imaging methods (Event related potentials - ERPs) & functional magnetic resonance imaging - fmri) have helped to overcome some of those limitations
Outline EEG & ERP methodology ERP evidence for early and late stage attentional modulations Multiple sub-systems view of attention Exogenous (stimulus driven) vs. Endogenous (voluntary) attentional allocation Sleep deprivation effects on attention
Offers excellent time resolution of neural events, but limited spatial resolution. Electroencephalography (EEG) Measures the electric potential differences across scalp arising from current flow within the head & brain.
Recording EEG - placement of electrodes
EEG signals arise primarily from post-synaptic activity (excitatory & inhibitory) Figures from Niedermeyer et al. (1999)
EEG signals represent the summed activity of large populations of synchronously firing neurons Figure from Speckmann and Elger (1999) Like holding a microphone over a stadium
Are all reliable neural events reflected in the EEG? Closed fields cancel each other out (e.g. midbrain nuclei) and are therefore invisible on the scalp Open field organizations (dendrites on one side, axon on other) summate. These include: most parts of cortex, parts of thalamus, cerebellum) Figures from Lorente de No. (1947) & Hubbard et al. (1969).
EEG properties of cortical patches are well described as current dipoles/dipole layers Neural dipoles => electromagnetic field changes throughout the head => current flow through head and scalp.
Dipolar nature of neural sources implies EEG signals arise from cortical gyrus sources Figure from Nunez (1995)
Event Related Potentials (ERPs) During stimulus processing, neural responses produce small scalp voltage changes embedded in the EEG trace. + time
Suppose these responses/voltage changes are consistent when processing similar stimuli in a similar way time
The problem is that this response is likely to be tiny with respect to the background noise ERPs are revealed by averaging the response across many trials
+ + The ERP reflects the average across-trial brain activity in response to the stimulus. C1 N1 Figures modified from (Pantev, 1995) P1 P3
Features and assumptions of averaging Activity reflects both signal and noise Signal: stimulus related processing Noise: tonic background activity related to ongoing processes (level of arousal, etc) The signal-related activity can be extracted because it is time-locked to the presentation of the stimulus we know exactly when it begins
ERP Components Segments of the ERP that covary with experimental manipulations Total ERP is an aggregate of numerous ERP components a peak or trough may represent the sum of several functional sources Components are quantified by peak amplitude/latency
Visual ERP components reflect different processing stages of the visual cortical pathway. Logothetis (2002)
C1 C1 reflects activity in primary visual cortex & is sensitive to stimulus properties. Logothetis (2002)
P1 P1 ~ activity in extrastriate visual cortex; sensitive to stimulus properties & information processing demands. Logothetis (2002)
N1/N2 ~ anterior lateral/ventral visual cortex. Sensitive to processing demands & stimulus meaning. N1 Logothetis (2002)
P2/P3 ~ posterior visual cortex. Sensitive to processing demands, stimulus meaning, & state of the organism. P3 Logothetis (2002)
ERP evidence for early selection Early ERP components in response to attended stimuli are larger than for unattended stimuli. Luck et al. (2000)
Spatial cueing ERP paradigms Exogenous (stimulus-driven) cueing Figure adapted from Luck et al. (1994)
P1: Neutral > Invalid Cost of attentional allocation in terms of stimulus detectability N1: Valid > Neutral Benefit of attentional allocation in terms of stimulus detectability Figure adapted from Luck et al. (1994)
Endogenous (voluntary attention) Posner (1980) cueing paradigm: Ss detected targets from rotated distractors - cues appeared for 950-1150 msec and targets for 125 msec. (Schnyer et al., unpublished data) Valid Trials - 80% Invalid Trials - 20%
2μv N1 225 ms P1 (Schnyer et al., unpublished data)
Object-Based Attentional Selection Attention can be directed to spatially overlapping objects & surfaces. Valdes-Sosa et al. (1998).
Larger ERP responses to attended vs. unattended overlapping objects. Valdes-Sosa et al. (1998).
ERP Evidence for Late Stage Selection Attentional Blink Decreased accuracy and P3 ERP with medium length delays btw successive targets. P3 reflect updates of working memory Attentional blink => interaction of attention & WM
Psychological Refractory Period Task Decreased accuracy and P3 ERP with medium length delays btw successive targets. No large changes in P3 response, but
Osman & Moore (1993) Lateralized readiness potential (LRP, reflective of motor processing) significantly delayed with short T1-T2 SOAs. => Interaction btw attention & response-related processing.
Multiple Subsystems View of Attention Early and late selection effects => attention operates across a variety of cognitive sub-systems: early sensory analysis object recognition working memory response selection & execution For a more extensive and recent review, see E.I. Kudsen, Fundamental Components of Attention, Annu. Rev. Neurosci. 2007. 30:57 78
Multiple Subsystems View of Attention Recent research has begun to examine: 1) the operation of attention across these different cognitive processing levels within single task paradigms 2) Different brain networks associated with attentional modulation across these processing levels
Attention Network Task (ANT) Fan et al. (2002; 2005) Functionally indexes attention networks involved in: 1) Alertness 2) sensory orienting 3) resolution of conflicting information
Fan et al. (2005)
Alerting Network Thalamus & frontal/parietal cortical sites
Orienting Network L/R superior parietal lobe, L precentral gyrus (near the FEFs)
Executive Control Network Anterior cingulate, L/R frontal areas, fusiform gyrus
The ANT task has been used to index EEG correlates of all three attentional networks (Fan et al., 2007).
All figures from Fan et al. (2007)
All figures from Fan et al. (2007)
Trujillo & Schnyer (in progress): Modified ANT Ss categorized target letter strings (consistent or inconsistent) under 3 cueing conditions (no cue, neutral cue, spatial cue). Exogenous and endogenous ANTs differed by nature of spatial cueing: EXO: cue location ENDO: cue orientation
1 Exogenous RTs Accuracy 0.98 0.96 0.94 0.92 0.9 Alerting effect: ns Orienting effect: Neutral Cue > Spatial Cue, p < 0.001 Conflict effect: Incon > Con, p < 0.001 0.88 spatial cue neutral cue no cue consistent inconsistent EXO ENDO 800 Endogenous RTs Alerting effect: No Cue > Neutral Cue, p < 0.001 Orienting effect: Neutral Cue > Spatial Cue, p < 0.001 RT (ms) 700 Conflict effect: Incon > Con, p < 0.001 600 spatial cue neutral cue no cue consistent inconsistent EXO ENDO
Exogenous ANT Target ERP Components Spatial Neutral No Cue Con Incon P1 N1 P3 - + μv
Endogenous ANT Target ERPs Components Spatial Neutral No Cue Con Incon P1 N1 P3 - + μv
Exogenous ANT Target ERPs Endogenous ANT Target ERPs N1 N1 P1 P3 P1 P3 Alerting, p < 0.05 Orienting, p < 0.08 Alerting, p < 0.004 Orienting, ns Alerting, p < 0.001 Orienting, p < ns Alerting, p < 0.002 Orienting, ns No P1 effects were significant.
Exogenous ANT Conflict ERPs Endogenous ANT Conflict ERPs N1 N1 P1 P3 P1 P3 No Conflict ERP effects were significant.
Exogenous ANT Cue ERPs Spatial Neutral No Cue P1 N1 P3 - + μv
Endogenous ANT Cue ERPs Spatial Neutral No Cue P1 N1 P3 - + μv
Exogenous ANT Cue ERPs Endogenous ANT Cue ERPs N1 N1 P1 P3 P1 P3 Neutral > Spatial, p < 0.001 Spatial > Neutral, p < 0.016 No P1 effects were significant. Spatial > Neutral, p < 0.036
Accuracy 1 0.99 0.98 0.97 0.96 0.95 0.94 0.93 Experiment 2 Changed task instruction to reduce conflict. Ss judged if center letter was same or different then flankers, rather than identity of center letter (Exp 1). 0.92 0.91 0.9 spatial cue neutral cue no cue consistent inconsistent 800 EXO ENDO Alerting, Orienting, & Conflict RT effects were all significant for both Exo & Endo tasks. RT (ms) 700 No accuracy differences. 600 spatial cue neutral cue no cue consistent inconsistent EXO ENDO
N1 Target ERPS P3 Target ERPS N1 & P3 Orienting effects were Significant for both Exo & Endo ANTs (ps < 0.001). All P1 and Conflict effects were non-significant.
The same N1 cue ERP effect was observed as in Exp 1. EXO: Neutral > Spatial; ENDO: Spatial > Neutral, ps < 0.04 The same ENDO P3 cue ERP effect was also observed: Spatial < Neutral, p < 0.003
Exogenous vs. Endogenous ANT Summary In general, Alerting, Orienting, and Conflict effects observed in RT data for both tasks. Orienting effects observed in Target-locked ERPs. No btw task differences. N1 Spatial vs. Neutral Cue-locked ERP differences observed for EXO & ENDO tasks P3 Spatial vs. Neutral Cue-locked ERP differences for ENDO.
Exogenous vs. Endogenous ANT Summary EXO Cue-locked ERP differences may reflect presentation of Spatial Cues in upper/lower visual fields vs. centrallypresented Neutral cue. ENDO Cue-locked ERP differences may reflect the taskrelated meaningfulness of the cues. We will further test this hypothesis with a new experiment using abstract symbolic cues devoid of spatial meaning.
The Influence of Sleep Deprivation on Attention Subjects: 9 West Point Cadets, 11 Ft. Hood soldiers. Sixty nine channels of electroencephalographic (EEG) signals recorded from subjects on two days, Day 1 (fresh) & Day 2 (fatigued), separated by 24 36 hours. Attention Network Task (ANT; Fan et al., 2002; 2005; 2007): Uses EEG signals to index brain networks involved in exogenous (externally driven) and endogenous (internally driven) allocation of selective spatial attention.
1 0.9 0.8 0.7 1000 900 WP/UT Accuracy 0.6 0.5 0.4 RT (ms) 800 0.3 0.2 0.1 WP/UT 700 0 spatial cue neutral cue no cue consistent inconsistent no response 600 spatial cue neutral cue no cue consistent inconsistent EXO DAY 1 EXO DAY 2 ENDO DAY 1 ENDO DAY 2 EXO DAY 1 EXO DAY 2 ENDO DAY 1 ENDO DAY 2 1 0.9 1000 FH 0.8 0.7 900 Accuracy 0.6 0.5 0.4 RT (ms) 800 0.3 0.2 FH 700 0.1 0 spatial cue neutral cue no cue consistent inconsistent no response EXO DAY 1 EXO DAY 2 ENDO DAY 1 ENDO DAY 2 600 spatial cue neutral cue no cue consistent inconsistent EXO DAY 1 EXO DAY 2 ENDO DAY 1 ENDO DAY 2
ANT Behavioral Results Summary 1) Lower accuracy & greater % of non-responses on Day 2 vs. Day 1, ps < 0.006. 2) Longer RTs on Day 2 vs. Day 1, ps < 0.001. 3) Expected effect of Spatial RTs < No Cue RTs for both tasks, ps < 0.001 Spatial RTs < Neutral RTs for Endo task only (p < 0.001). Congruency x Group (p < 0.007) interaction for Endo task (p < 0.001) WPs RTs < FH RTs for Con & Incon, with larger difference for Incon trials. Day x Cue Type interaction for Endo task (p < 0.022) Day 1 Neutral RTs < No Cue RTs, Day 2 Neutral RTs > No Cue. 4) Expected effect of Con RTs < Incon RTs for both tasks, ps < 0.001
Exogenous ANT Day 1 Spatial Neutral No Cue Con Incon P1 N1 P3 - + μv
Exogenous ANT Day 2 Spatial Neutral No Cue Con Incon P1 N1 P3 - + μv
Endogenous ANT Day 1 Spatial Neutral No Cue Con Incon P1 N1 P3 - + μv
Endogenous ANT Day 2 Spatial Neutral No Cue Con Incon P1 N1 P3 - + μv
Endogenous ANT _ Targets Day 1 Day 2 O1 O1 N1: Day 1 > Day 2, p < 0.02; Spatial > Neutral/No Cue, ps < 0.002 P3: Spatial < Neutral/No Cue, ps < 0.03
Exogenous ANT _ Targets Day 1 Day 2 O1 O1 P1: Spatial > Neutral/No Cue, ps < 0.02 N1: Spatial > Neutral/No Cue, ps < 0.001 P3: Spatial > Neutral/No Cue & Neutral > No Cue, ps < 0.013
P3: Cue Type x Group Interaction, p < 0.031 WP/UT FH Spatial Neutral No Cue
Exogenous ANT _ Congruency Day 1 Day 2 O1 O1 N1: Day 1 > Day 2, p < 0.022
Endogenous ANT _ Congruency Day 1 Day 2 O1 O1 N1: Day x Congruency interaction, p < 0.029
N1: Day x Congruency Interaction, p < 0.029 Con Incon
The Influence of Sleep Deprivation on Attention: Preliminary Conclusions The present pattern of behavioral & ERP results suggest that fatigue can affect even the low-level ability to attend to objects/locations in visual space. Sleep loss seems to affect endogenous (voluntary) attention to a greater degree than exogenous (stimulus-based) attention.
Thanks For Your Attention! Mora and Carmen (1995)