Motor evoked potentials reflect changes in rapid inhibitory control during serial ordering.
2017, Behmer, L. P., Crump, M. J. C, Jantzen, K. J., Martinez, S., Walls, R., Amir-Brownstein, E., Jaye, A., Leytze, M., Lucier, K. Cognitive Neuroscience Society.
Abstract: Theories of serial ordering assume all responses in a sequence are activated in parallel and held in a buffer for execution. An inhibition or timing process impedes responses in a graded fashion with earlier more active responses executed before more inhibited later responses. There is no direct evidence in humans that planned responses are inhibited as a function of serial order. The necessary evidence could be provided by a response activation “thermometer” measuring whether the current “temperature” or activation level is graded by position across all responses in the buffer. We used transcranial magnetic stimulation to probe the level of excitation for flexion of the right index finger (first dorsal interosius muscle mucle) during typing. Motor evoked potentials (MEPs) were recorded at the onset of typing of 5-letter words and nonwords. A single letter typed by the right index finger varied across letter positions one to five. The amplitude of the MEP when the right index finger was used in letter positions one and two reflected the relatively active state of the FDI. MEP amplitude decreased monotonically with further increases in position. Interestingly we also found smaller MEPs when the right index finger was the first rather than second response, indicative of a rapid deactivation at the completion of a key press. This is the most direct human evidence to date corroborating inhibition/timing theories, showing that completed responses are rapidly deactivated and future responses are activated in a graded fashion as a function of serial position.
Learning to Selectively Ignore Information
2017, Brosowsky, N. B., & Crump, M. J. C. Eastern Psychological Association, Boston.
Abstract: The goal of the current study was to use traditional learning procedures to examine how learning processes might participate in the adjustment of selective attention processing. Across three experiments, we found that participants could learn to ignore a particular word in a Stroop task, eliminating the traditional interference effect. However, in all three experiments the interference effect re-appeared when the trained items were intermixed with untrained items, demonstrating a consistent failure of transfer.
Cognitive Control for Action Sequencing
2017, Behmer, L. P., & Crump, M. J. C. Eastern Psychological Association, Boston.
Abstract: Here, we present evidence integrating cognitive control principles and neural networks into an account of action sequencing. While recording continuous EEG during typing tasks, we observed increased anterior cingulate theta-power as a function of co-activation of competing keypresses responses and increased frontal theta-power during conditions where proactive control was required, such as while explicitly monitoring typing output. These findings support the possibility that the neural mechanisms driving cognitive control also mediate control of actions.
Theta-Event Related Synchrony (ERS) as an Index of Outer- Loop Control During Typing: Neural Evidence Supporting the Two-Loop Theory of Typing.
2016, Behmer, L. P., & Crump, M. J. C. Psychonomics, Boston
Abstract: Skilled typists accurately type 4-5 letters per second, yet have poor declarative memory of key locations. This suggests typing may be controlled by two hierarchically nested control loops that are informationally encapsulated from one another. We present behavioral and EEG [frontal midline theta-event related synchrony (ERS)] evidence that examines the development of hierarchical control across trials. Participants typed a series of 7-letter words, ten times in a row, on a QWERTY keyboard and a keyboard where letter mappings had been jumbled. We found that typists were slower and showed increased theta-ERS for jumbled early compared to jumbled late, QWERTY early, and QWERTY late trials. Although typing speeds improved for jumbled late trials, theta- ERS was still greater than baseline. These findings suggest that theta-ERS may index levels of outer-loop control during typing, and even though inner loop control was rapidly established when typing on the jumbled keyboard, outer-loop engagement remained robust.
Long(er)-Term Item-Specific Gratton Effects.
2016, Brosowsky, N. P., & Crump, M. J. C. Psychonomics, Boston
Abstract: Gratton effects refer to the finding in selective attention tasks (e.g., Stroop, Flanker) that interference effects (incongruent minus congruent RTs) are smaller on trials immediately following an incongruent than congruent trial. Gratton effects are commonly explained by transient or reactive conflict-induced shifts in attentional weights on trial n-1 that carry forward to influence trial n. Separately from this short-term influence, episodic memory traces may also bind together attentional control settings associated with features of particular items that could allow for longer-term cue-driven reinstatement of prior attentional settings. In three related experiments, participants completed a color-identification flanker task. Colors were presented on different images, and specific images were paired with congruent or incongruent trial types. We found that re-presenting an image 4-8 trials later modulated congruency effects on an item-specific, and longer- term basis than the traditional Gratton effect.
Context-Specific Proportion Congruent Effects for Frequency Unbiased Items: Novel Extensions and Analyses of Reproducibility.
2016, Crump, M. J. C., & Brosowsky, N. P., International Psychonomics, Granada
Abstract: Crump & Milliken (2009) showed that Stroop effects can be modulated by context-specific cues associated with different levels of proportion congruent, even for items that appeared equally frequently in each context. This result has important theoretical implications, because it ruled out frequency-driven learning explanations of context-specific proportion congruent (CSPC) effects, and left open the possibility that cue-driven retrieval process can reinstate attentional control setting in a rapid online fashion. The purpose of the present work was to address reproducibility concerns that have been raised about this finding. We conducted several replications and novels extensions of the Crump & Milliken procedures using Amazon’s mechanical Turk in both Stroop and flanker tasks. We successfully replicated the major findings from prior work, and conducted new analyses of the role of sequential influences on CSPC effects that showed a critical role for context repetitions in CSPC flanker effects, but not in CSPC Stroop effects. We also provide new monte-carlo simulation analyses to estimate reproducibility of the phenomena that show important limitations on these designs for measuring contextual control.
How Typists Talk to Their Fingers: Evidence for Word-Level Verbal Control of Skilled Action Sequencing.
2016, Brosowsky, N. P., Behmer, L., Crump, M. J. C., International Psychonomics, Granada
Abstract: Hierarchical theories of action planning in skilled domains like typing (Logan & Crump, 2011) have proposed a distinction between inner and outer loop control processes that represent higher-order planning (outer loop) processes and lower-level motor execution (inner loop) processes. Specifically, the outer loop control processes retrieve words as plans, and serves word-level representations to the inner loop, which then translates them to individual keystrokes at the letter level. A critical implication of the dual-loop theory is that typists rely on verbal processes to control serial-ordering between finger movements. The current study tests the role of verbal processes in controlling fluent typing in three tasks manipulating verbal processing during typing. In Experiments 1 and 2, we demonstrate that typists type faster if they speak the words rather than letters. Experiments 3 and 4 show that typing performance is worse under conditions where verbal processing is disrupted. In Experiment 3, typing was disrupted under several conditions of verbal suppression and in Experiment 4, under conditions of delayed auditory feedback. Taken together, these results provide supporting evidence for the dual-loop hierarchical control theory and suggests a role for verbal processes in the control of fluent typing.
Try not to think about what you?re doing: Event-related desynchronization (ERD) in the SMA and event-related synchronization (ERS) in the frontal lobe reveals a dissociation of hierarchically controlled outer and inner loop processes during typing.
2016, Behmer, L. P., & Crump, M. J. C., Cognitive Neuroscience Society. New York, NY.
Abstract: Attention to the details of actions disrupts performance. For example, forcing people to type letters of a word assigned to one hand compared to all letters leads to slower RTs (Logan & Crump, 2009). Given that people are generally fast at normal typing, this suggests that there may be a two-loop control system for skilled actions in which an outer loop is involved in planning, and an inner loop executes kinematic actions, independent of outer loop oversight. Previous EEG studies have focused only on measuring parallel response ordering by the inner loop during normal typing (Logan, Miller, & Strayer, 2011; Pinet et al., 2015). Here, we present behavioral and EEG evidence dissociating outer and inner loop processes. Participants typed four letter words under conditions that varied whether the outer loop closely monitored the details of action execution. Explicit monitoring required that participants type only the letters exclusively assigned to the left or right hand. Normal typing required considerably less monitoring; participants typed all of the letters of the word with both hands. RTs were significantly faster for normal versus monitored typing. Also, alpha-ERD in bilateral clusters of the SMA at about 500 ms prior to the first keystroke was significantly greater during normal versus monitored typing. Finally, during the typing interval, alpha-ERS in the frontal cluster was greater during monitored typing. This suggests outer loop monitoring of the output of the inner loop leads to deficits in performance and an increase in premotor and higher cognitive processing.
The dynamics of best laid action plans: Tracing response scheduling in skilled typing.
2016, Behmer, L. P. & Crump, M. J. C., Eastern Psychological Association, New York, NY.
Abstract: This talk presents evidence for predictions made by computational models about parallel and serial activation of action sequences during skilled motor performance. While employing a go-signal procedure that forced typists to occasionally stop and type a previous or future letter, we observed that future responses displayed graded activation states consistent with parallel models of activation. These findings support competitive queuing models employing a winner-take-all mechanism for selecting the most active element over other inhibited elements.
Contextual control over stimulus-response sets.
2016, Brosowsky, N. P. & Crump, M. J. C., Eastern Psychological Association, New York, NY.
Abstract: Context has been shown to cue the automatic reinstatement of various psychological processes across domains. The current study investigates context-dependency in a previously unexamined aspect of cognitive control: the creation and execution of stimulus-response sets. Specifically, we investigate whether response subsets can become associated with, and retrieved by contextual cues. Over three choice reaction time experiments we found no evidence for contextual control over stimulus-response sets.
Typists develop sensitivity to the statistical frequencies of likely N-grams.
2015, Behmer, L.P. & Crump, M. J. C, Eastern Psychological Association, Philadelphia.
Abstract: We investigate how serial ordering abilities develop with practice in the context of skilled typing. Using a large N online study we show that poor typists’ individual keystroke typing times were strongly correlated with single letter frequency (faster when typing high-frequency single letters), whereas expert typists were correlated with bigram and trigram frequencies. These data show that skill acquisition for action sequencing relies on a learning process sensitive to statistical structure of produced actions.
Evidence for generalization of context-specific control in a flanker task
2015, Brosowsky, N. P., & Crump, M. J. C., Eastern Psychological Association, Philadelphia.
Abstract: The present work investigates stimulus-driven setting and adjusting of attention filtering demands. In a flanker task, the attentional sets applied to one set of congruent and incongruent items in particular location contexts were found to be applied in a general fashion to control processing of another set of frequency unbiased items presented in the same contexts. The results license further discussion of how learning and memory processes participate in attentional control.
Contextual control of attentional sampling: Exploring the role of volition.
2015, Brosowsky, N. P., & Crump, M. J. C. Psychonomics, Chicago, IL.
Abstract: Contextual properties of the environment, through learning and memory processes, can rapidly and involuntarily adjust attentional control (Chun, 2000; Corballis & Gratton, 2003; Crump, Gong, & Milliken, 2006; Crump, Vaquero, & Milliken, 2008). The current study explores the obligatory nature of contextual cueing, specifically examining the role of volition in acquiring context-dependent attentional control in a bi-dimensional (colors vs. letters) stimulus sampling task. Subjects viewed briefly presented arrays of letters and colors presented above or below fixation, and identified specific stimuli given a task-cue associated to each location. Experiments 1 and 2 showed no contextual control over priority for sampling either dimension. Experiment 3 showed context-dependent prioritization when participants were instructed to voluntarily deploy different sampling strategies between contexts. Experiment 4 showed that context-specific sampling strategies also depends partly on non-voluntary learning processes.
False Memory for Previously Tested Items: Investigating Test-Induced Associative and Repetition Priming.
2014, Crump, M. J. C., Blecher, M., & Jamieson, R., Psychonomics.
Abstract: Studying thematically related words produces false memories for non-studied theme words (i.e., DRM effect). Test-induced priming (TIP) is known to increase false-recognition rates when source monitoring is impaired. Our experiments establish the reliability of associative and repetition TIP effects for non-studied lists, which have been inconsistently shown in the literature. In several online experiments, participants encoded words never presented at test. At test, thematically related new words were presented as lures before or after critical category lures. Participants showed reliable false memories for critical lures (E1). TIP effects remained when encoding time increased (E2), when warnings were given (E3 & E4), and when old words from the study phase were included in the test (E6). False-recognition rates were higher for lures primed by repetition rather than association during test (E5 & E6). The role of source- monitoring processes during retrieval in mediating false- recognition from TIP is discussed.
Stimulus-Driven Control of Explicitly Induced Attention Filtering Demands.
2013, Crump, M. J. C., Psychonomics.
Abstract: Demonstrations of item/context-specific proportion congruent (PC) effects suggest that early stimulus processing can rapidly control attention filtering (see Bugg & Crump, 2012). Yet, PC manipulations are typically confounded with item-frequency, and PC effects may be driven by a learning process sensitive to item-frequencies. The present work examines whether explicitly induced attention filtering demands can transition from strategic to stimulus-driven control without manipulating PC or biasing item-frequencies. Several web-based flanker studies were conducted via Amazon Mechanical Turk. Each involved a 50/50 PC design. Attention filtering demands were induced by a 1-back memory task following each trial prompting recall of the last target or flanking letter. Blocking recall demands produced list-wide PC-like effects with larger flanker effects for flanker than target recall conditions. Mixing recall demands and associating them with particular stimulus-cues (location, color, letters, font) produced item/context-specific PC-like effects, and show novel unambiguous evidence for stimulus-driven control of attention filtering.
Prevention or Cure? Error Detection and Correction in Skilled Typists.
2012, Logan, G. D., & Crump, M. J. C., Psychonomics.
Abstract: Error detection serves two different functions: prevention and cure. Prevention engages post-error slowing to reduce future errors. Cure engages processes that correct the error. Thus, prevention predicts post-error slowing and cure does not. Skilled typists prefer cure to prevention: In three experiments, we found post-error slowing only when correction was disallowed, and the slowing persisted only for a few keystrokes. Words that followed errors were typed as quickly as words that preceded errors. We found no post-error slowing when typists were allowed to correct errors. They corrected errors as soon as they detected them and then resumed typing at normal rates. Keystrokes that fell between the error and the error detection response were actually faster than keystrokes before the error, suggesting post-error speeding. These findings led us to propose the novel hypothesis that post-error slowing reflects the inhibition of pre-potent tendencies to correct mistakes.
An Instance-Based Account of Retrospective Revaluation.
2012, RANDALL K. JAMIESON and CHRISSY M. CHUBALA, University of Manitoba, SAMUEL D. HANNAH, University of Saskatchewan, MATTHEW J. C. CRUMP, Brooklyn College, CUNY, Psychonomics.
Abstract: After learning that a compound cue AB predicts an outcome X, learning that A alone predicts X forces a deflation in the predictive validity of B—a phenomenon called retrospective revaluation. We model retrospective revaluation using an instance model of associative learning based on the MINERVA 2 model of human memory. According to the model, each experience is stored to memory as a unique trace and response strength is determined by the similarity between the target outcome and the information that a cue retrieves from memory. We show that the model accommodates retrospective revaluation at the first and higher orders of association and we argue that associative learning is consistent with an instance-based approach to learning.
Can Internet Data be Trusted? Validating Mechanical Turk for Cognitive Research.
2012, MATTHEW J. C. CRUMP, Brooklyn College, CUNY, TODD M. GURECKIS and JOHN V. MCDONNELL, New York University, Psychonomics.
Abstract: Amazon Mechanical Turk is an online crowd sourcing service where anonymous online workers complete web-based tasks for small sums of money. There are many workers willing to perform tasks, and an experiment with n=100 can easily be conducted over night. Amazon Turk has become increasingly popular in the decision making literature, thereby partly demonstrating the viability of the service for running psychology experiments. However, the service has not been validated for cognitive research in general which typically requires multi-trial designs and millisecond accuracy for response recording and stimulus presentation. The service was validated by replicating several classic cognitive tasks including: Stroop, Task-switching, Flanker, Simon, Visual cuing, Attentional blink, Subliminal priming, and Category learning tasks. The replications were mostly successful and validated the approach. A wide-variety of behavioral research can be conducted in a rapid, cost- effective manner using Amazon Mechanical Turk.
Hitting the Wall in Skilled Typing: Speed/Accuracy Tradeoffs and Serial Processing.
2011, MATTHEW JC CRUMP, Brooklyn College of CUNY, GORDON D. LOGAN, Vanderbilt University, Psychonomics.
Abstract: Two experiments investigated speed-accuracy tradeoff functions in a continuous typing task. Experiment 1 had typists adjust their typing rate by 20% faster or slower than their normal rate by following a metronome, a speedometer, or a moving color cue. Accuracy was minimally influenced by changes in typing speed. Experiment 2 parametrically varied typing rate in 10% steps by having typists attempt to type up to 90% faster than their normal rate. Typists were able to maintain low error rates over large changes in speed. Analysis of the distribution of errors across speed stress shows increases to the number of letter omissions and substitutions, but not transpositions. We suggest that typists manage the speed- accuracy tradeoff by attempting to preserve serially-ordered responses.