Parsing under pressure: The role of performance pressure in cognitive control and syntactic ambiguity resolution E.K. Hussey, S.E. Teubner-Rhodes, M.R. Dougherty, & J.M. Novick Neuroscience and Cognitive Science Program, Department of Psychology University of Maryland, College Park Half of the par,cipants were told that all tasks were assessments of intelligence. The Role of Cogni-ve Control in Sentence Processing  Psycholinguis,c studies 1,2 emphasize the importance of cogni,ve control during sentence processing, especially when readers/listeners must override early, incorrect interpreta,ons due to late‐arriving conflic,ng input.  Extreme popula,ons, including young children 3,4 and pa,ents 5,6 with damage to leH lateral prefrontal cortex, have difficulty resolving incompa,ble representa,ons during cogni,ve control tasks (e.g., Stroop) 2,5 . This general cogni,ve control difficulty is also reflected in garden‐path recovery 2,4,5 . Cogni-ve Control and Mo-va-on  Recent research demonstrates that performance on cogni,ve control tasks (e.g., Stroop) is affected by mo,va,onal state 7 :  Approach State: mo,vated by reward‐associated cues to aSain posi,ve outcomes  Avoidance State: mo,vated by threat‐associated cues to withdraw from aversive outcomes  Similarly, performance on cogni,ve control tasks decline when individuals' mo,va,onal states change, such as when they are pressured to perform well 8 . Current Predic-ons  If cogni,ve control helps readers implement reanalysis during sentence processing, then manipula,ng the demands for control by inducing pressure should result in measurable differences during sentence processing.  Specifically, readers under performance pressure might show greater difficulty revising early interpreta,on commitments.  Individuals under pressure (as compared to controls) show increased processing difficulty while reading temporarily ambiguous sentences.  Notably, this difficulty is confined to sentence regions where the input conflicts with one’s developing interpreta,on, triggering controlled revision processes.  Stroop performance suggests that these findings are not due to baseline cogni,ve control differences between the groups.  The current manipula,on revealed only online effects of pressure during syntac,c ambiguity resolu,on. Future studies should focus on offline lingering garden path effects, perhaps by using a moving window paradigm.  These findings extend research integra,ng mo,va,on and cogni,ve control: Real‐,me sentence processing is influenced by pressure under selec,ve condi,ons, namely when control must be ini,ated to recover from misinterpreta,on.  Future work will involve beSer measures of mo,va,on, state/trait verifica,on, individual difference components, as well as studies of spoken comprehension and lexical access. Background & Hypotheses Methods Results No Group Differences on Stroop Prior to Pressure Manipula-on Normalized Stroop Cost Normalized Stroop Cost = Error Bars = SEs ehussey@psyc.umd.edu, steuberr@psyc.umd.edu mdougherty@psyc.umd.edu, jnovick1@umd.edu N = 21 (11 Control; 10 Pressured) 1. Novick, Trueswell, & Thompson‐Schill. (2005). Cogni,ve control and parsing: Reexamining the role of Broca’s area in sentence comprehension. CABN 2. January, Trueswell, Thompson‐Schill. (2009). Co‐localiza,on of Stroop and Syntac,c Ambiguity Resolu,on in Broca’s Area: Implica,ons for the Neural Basis of Sentence Processing. J. Cog. Neuro. 3. Davidson, Amso, Cruess Anderson, & Diamond, (2006). Development of cogni,ve control and execu,ve func,ons from 4 to 13 years: Evidence from manipula,ons of memory, inhibi,on, and task switching. Neuropsyc. 4. Trueswell, Sekerina, HIll, & Logrip. (1999). The kindergarten‐path effect: Studying online sentence processing in children. Cogni3on. 5. Novick, Kan, Trueswell, & Thompson‐Schill. (2009). A case for conflict across mul,ple domains: Memory and language impairments following damage to ventrolateral prefrontal cortex. Cog. Neuropsych. 6. Hamilton & Mar,n. (2005). Dissocia,ons among tasks involving inhibi,on: A single case‐study. CABN. 7. Prabhakaran, Kraemer, & Thompson‐Schill. (2009). Personality traits predict cogni,ve control ability. Poster presented at Annual Psychonomic Society Mee3ng. 8. Beilock & Carr. (2005). When high‐powered people fail: Working memory and “choking under pressure” in math. Psych Science. 9. Chris,anson, Hollingworth, Halliwell, & Ferreira. (2001). Thema,c roles assigned along the garden path linger. Cog Psych. Summary & Conclusions References First‐Pass Reading: No Differences Between Pressured and Non‐pressured Readers Pressured Control while Anna dressed. The baby spit up that was cute and cuddly on the bed spit up on the bed. While Anna dressed the baby that was cute and cuddly Second‐Pass Reading: Pressured Readers Spend Significantly More Time Re‐reading Ambiguous Sentences but only in the Disambigua-ng Region! Pressured Readers “Back‐up” Significantly More Into Regions Where Revision is Needed! The baby spit up While Anna dressed the baby that was cute and cuddly spit up that was cute and cuddly on the bed Correct Responses Only Unambiguous Sentences Ambiguous Sentences Unambiguous Sentences Ambiguous Sentences Unambiguous Sentences Ambiguous Sentences spit up on the bed. While Anna dressed the baby that was cute and cuddly while Anna dressed. The baby spit up that was cute and cuddly on the bed Correct Responses Only Correct Responses Only No Effect of Pressure on Off‐line Comprehension Measure Ambiguous Unambiguous Control Pressured Propor-on Correct Control Pressured Pressured Control Pressured Control Red Green Blue Yellow STROOP Non‐Parsing Cogni-ve Control Task Red! Blue! Congruent Trial Incongruent Trial Stroop Task Pressure Manipula-on Sentence Processing Tasks While Anna dressed the baby that was cute and cuddly spit up on the bed. LINGUISTIC MATERIALS Parsing Cogni-ve Control Task Reflexive Absolute Transi3ve Verb Sentence Construc3ons 9 Did Anna dress herself? Ambiguous Pressured Group Control Group Reading Time (ms) Probability of Regressing Into Region X Reading Time (ms) Probability of Regressing Into Region X Eye movements were recorded during reading using an Eyelink 1000. The baby that was cute and cuddly spit up on the bed while Anna dressed. Did Anna dress herself? Unambiguous Pressure Manipula-on Apparatus Reading Time (ms) Reading Time (ms)