P031 - “One task to rule them all: Disentangling episodic and semantic memory formation”
1Universität Tübingen, Deutschland; 2Philipps-Universität, Marburg, Deutschland
Introduction. Systems memory consolidation postulates two brain systems for declarative memory coding different aspects of a memory, with the hippocampal system storing episodic and detailed information and the neocortex extracting regularities to generate concepts/schemata (Marr, 1970). Experimental paradigms are usually tailored to engage only one system, making it impossible to directly compare results. Therefore, we have designed a memory task that allows for identical visual stimulation while targeting either the episodic (EPI) or semantic system (SEM) via different instructions. Here we present preliminary data evaluating this novel paradigm.
Methods. During learning, 34 participants are presented with unknown stimuli (“aliens”) in different contexts (“planets”). While every alien lives on one planet and has a unique appearance, some also share a similar shape/color scheme. Two experimental groups differ in the instructions on how to encode the stimuli with either having to categorize them into groups of aliens with shared features (SEM) or encode the stimuli based on individual features including the context (EPI). 24h later the participants perform two memory tests: A categorization task targeting the semantic system and an episodic recognition task.
Results. While there was no difference between conditions in performance levels and subjective task difficulty for the encoding task, SEM performed significantly better identifying new category members but worse at recognizing correct stimulus-context pairings.
Discussion. Preliminary results indicate that this paradigm induces differential engagement of episodic and sematic memory systems only by varying instructions, opening up new possibilities for neuroimaging to observe parallel memory formation and investigate systems interaction.
P032 - Decoding Human Memory Replay During Rest
1Zentralinstitut für seelische Gesundheit, Mannheim, Deutschland; 2Universität Tübingen, Deutschland
Animal research suggests replay of memory traces during rest and sleep improves performance, but this research relies on highly invasive methods and only has access to simple behavior. Similarly, quantifying replay in humans has proven difficult, and up to this point, no study exists that is able to detect endogenous human memory replay during sleep. In this proof-of-principle study, we applied a recently developed method in healthy humans to record item-level human replay events and thereby uncover the processes occurring during quiet rest. In subsequent studies we hope to further develop this method to be able apply it to human sleep.
In this study we attempted to detect sequential memory replay in humans during rest. In a first step, sixteen items were presented multiple times in a pseudorandom order while brain activity (MEG) was recorded in order to extract the representational brain state activity for these items. Machine learning classifiers were trained to decode the brain states belonging to each item. In a second step, the participants learned an ordering of the previously presented items. Subsequently, in a resting state condition, we were able to detect preliminary evidence for time-compressed replay of the learned items. Our analysis confirmed previous findings, that items are replayed with a time-lag of around 40-50 milliseconds between individual items.
P033 - Dynamics of nonlinguistic statistical learning: From neural entrainment to the emergence of explicit knowledge
1University of Tübingen, Germany; 2Western University, London, ON, Canada; 3University of Trento, Italy; 4Northwestern University, Evanston, IL, USA
Statistical learning, which describes the ability to detect patterns in the environment, plays an important role in many aspects of cognition. Up to now, there is little knowledge about the dynamics of this implicit form of learning and its relation to explicit knowledge. In the present study, 24 healthy volunteers were exposed to an auditory nonlinguistic statistical learning paradigm while their brain activity was measured with magnetoencephalography. The stimulation paradigm was composed of 12 pure sinusoidal tones, arranged in one structured – containing repeating tone triplets – and one random sequence. Statistical learning was quantified by neural entrainment to the embedded tone triplets. When exposed to the structured sequence, participants showed strong neural entrainment to the embedded triplet pattern, which was present already after a few triplet repetitions. Source reconstruction revealed that in addition to brain areas included in the auditory processing hierarchy, the left pre-central gyrus plays a role in this form of statistical learning. Furthermore, neural entrainment to the triplet patterns increased over time and its overall strength predicted participants’ subsequently tested explicit knowledge. This suggests that there is a systematic relationship between neural entrainment and explicit learning of triplet structures. While participants had some difficulty with explicitly expressing this kind of nonlinguistic learning, the measured neural entrainment and its temporal dynamics reflected a robust, implicit learning of underlying patterns.
P034 - Effects of contingency awareness on aversive and appetitive conditioning as revealed by pupillometry
Universität Siegen, Deutschland
Evidence regarding unaware differential conditioning in humans is mixed. Moreover, while the very existence of implicit fear conditioning has recently been challenged, even less is known about effects of contingency awareness on appetitive conditioning. Phasic pupil dilation responses (PDR) might yield a promising readout that is potentially more sensitive to implicit learning than skin conductance responses (SCR) or other more established measures. Thus, the present study varied the type of contingency instructions (between) as well as valence of the unconditioned stimulus (UCS) (within participants).
After being randomized to either a fully instructed or non-instructed group, participants underwent a differential conditioning procedure involving perceptually matched grating stimuli varying in orientation (angle) only. One conditioned stimulus (CS+app) was followed by a noise signaling monetary reward (1€) in 13 of 20 trials, whereas another one (CS+avs) was associated with a mild electric shock. A third visual stimulus (CS-) was never paired with either aversive or appetitive UCS. PDR and SCR were recorded during both acquisition and extinction training.
SCR results demonstrated successful differential fear conditioning in the group of instructed participants. In the subgroup of non-instructed participants who remained unaware of the CS-UCS association, no effect on SCR, yet differential modulation of early pupil responsivity by appetitive cues emerged. However, this effect was diminished during later stages of acquisition, suggesting reduced stimulus discrimination.
Our findings add to the growing body of research utilizing PDR as an index of differential conditioning, indicating that pupillometry might be better suited to track implicit conditioning than SCR.
P035 - Electrophysiological correlates of sleep-associated memory consolidation in pre-school children
Max-Planck-Institut für Bildungsforschung, Deutschland
Introduction: In adults, the precise temporal coordination of fast spindles (~12-16 Hz) and slow oscillations (< 1 Hz) is considered a key mediator of sleep-associated memory consolidation. However, across development, spindles and slow oscillations change considerably. Thus, it remains elusive whether the same mechanisms of sleep-associated memory consolidation as identified in adults are comparably functional in children.
Methods: Here, we characterise slow (~9-12 Hz) and fast spindles leveraging an individualised rhythm detection approach and their temporal modulation during slow oscillations using time-frequency analyses and peri-event time histograms in 24 5- to 6-year-olds. Further, we investigate whether spindles and their modulation during slow oscillations are associated with behavioural indicators of declarative memory consolidation.
Results: We reliably identify an endogenous, development-specific fast spindle type, though nested in the adult-like slow spindle frequency range, along with a dominant slow spindle type. Further, coupling analyses indicate the presence of fast spindle-slow oscillation coupling already in pre-school children – though weaker and less precise than expected from adult research. While we do not find evidence for a critical contribution of the pattern of fast spindle modulation during slow oscillations for memory consolidation, crucially, robust regressions show that slow and fast spindles are differentially related to memory consolidation of items of varying encoding quality.
Discussion: Our results reveal two functionally relevant spindle types in pre-school children despite not fully matured spindle–slow oscillation coupling. However, it remains an open question what renders spindle-slow oscillation coupling more precise and pronounced and how this affects memory consolidation across development.
P036 - Encoding associative memories based on feedback – an fMRI study
1Otto-von-Guericke-Universität Magdeburg, Deutschland; 2Center for Behavioral Brain Sciences, Magdeburg, Deutschland
Introduction: The ability to update memory representations by (re)learning after recognizing missing or wrong memories is central for adaptive performance in everyday life. However, there has not been much neurophysiological research about how performance monitoring may improve the formation of long-term memories.
Methods: We conducted an fMRI study on 30 young participants (15 female) using a feedback-guided associative learning paradigm, in which unknown faces had to be matched with gabor patches of eight different orientations. On each trial, participants judged the confidence in their choice, received feedback on the correctness and were presented with the correct stimulus combination to encode correct associations for following trials. Successful feedback-based encoding was defined as a corrected error, i.e., correct reproduction of the face-orientation association on the next trial.
Results: After choosing wrong face-orientation associations, negative feedback was associated with increased BOLD responses in the anterior insula, posterior medial frontal cortex (pMFC) and frontoparietal network. Positive feedback was accompanied by effects among ventral striatum and ventromedial prefrontal cortex. Successful encoding based on negative feedback – later corrected compared to repeated errors – showed an overlap with negative feedback particularly in pMFC, supplementary motor area, dorsolateral and inferior frontal regions. In addition, signal increases in the anterior hippocampal formation were found.
Interpretation: The current results suggest that regions associated with negative feedback may guide successful encoding after previous mistakes by recruiting hippocampal activity. Further analyses will investigate feedback-dependent functional connectivity of these regions.
P037 - Hippocampo-frontal multivoxel patterns are modulated by prefrontal stimulation and motor sequence learning
1Department of Movement Sciences, Motor Control and Neural Plasticity Research Group, KU Leuven, Belgium; 2Department of Health and Kinesiology, College of Health, University of Utah, USA; 3Department of Clinical Sciences, College of Health and Life Sciences, Brunel University London, United Kingdom; 4Brain Imaging and Neural Dynamics Research Group, IRCCS San Camillo Hospital, Italy; 5Institute of Neuroscience and Psychology, University of Glasgow, United Kingdom
Introduction: Motor sequence learning (MSL) is supported by dynamical interactions between striato- and hippocampo-cortical networks mediated by the prefrontal cortex. We examine whether multivoxel patterns in the dorsolateral prefrontal cortex (DLPFC) and hippocampus can be modulated via MSL and theta-burst-stimulation (TBS) of the DLPFC.
Methods: 19 participants received DLPFC-TBS before being scanned on a motor task in a 2x2-within-subject design [stimulation(iTBS/cTBS) by task(sequential/random)].
Resting-state (RS) scans were acquired before stimulation and after task performance.
For each ROI (DLPFC /hippocampus) and MRI session (RS pre- and post-stimulation/learning; early and late task), activity patterns were estimated by computing correlations among the n voxels extracted from each ROI. Pearson’s correlations were computed between each of n BOLD-fMRI voxel time courses, yielding an n-by-n matrix (multivoxel-correlation-structure, MVCS, Tambini&Davachi(2013)). Similarity of the multivoxel patterns between two MRI sessions were computed as the correlation between the MVCS matrices and entered into 2 (task) by 2 (stimulation) repeated-measures ANOVAs.
Results: DLPFC patterns were more similar between the early and late task stages in the random as compared to the sequence task (task effect).
Hippocampal patterns at rest were more similar after the random than the sequence task (task effect) and in the iTBS as compared to the cTBS conditions (stimulation effect).
Conclusions: Our results show that MSL disturbed patterns of the DLPFC during task practice and of the hippocampus during post-learning rest. Importantly, prefrontal stimulation affected hippocampal patterns during post-learning/stimulation rest. Collectively, these data provide the first evidence that both MSL and stimulation modulate hippocampo-frontal multivoxel patterns.
P038 - How sleep balances cortical circuit activity
1Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, 72076 Tübingen, Germany.; 2Center for Integrative Neuroscience, University of Tübingen, 72076 Tübingen, Germany; 3German Center for Diabetes Research (DZD), Institute for Diabetes Research & Metabolic Diseases of the Helmholtz Center Munich at the University Tübingen (IDM), 72076 Tübingen, Germany
Sleep shapes cortical network activity, contributing to a global homeostatic down-regulation of excitability, while in selected networks excitability is maintained or even up-regulated as a consequence of memory consolidation.
Here we used two-photon calcium imaging of cortical layer 2/3 neurons in sleeping mice to examine how these seemingly opposing dynamics are balanced in cortical networks.
During slow-wave sleep (SWS) epochs, mean activity of excitatory pyramidal (Pyr) cells was decreasing. Simultaneously, however, variance in Pyr population activity was increasing, contradicting a homogenous down-regulation of network activity. Indeed, we identified a subpopulation of Pyr cells distinctly upregulating activity during SWS, which were highly active during sleep spindles known to support mnemonic processing. REM epochs succeeding SWS produced a general down-regulation of Pyr cells, including spindle-active Pyr cells, which persisted into following stages of sleep and wakefulness. Parvalbumin-positive inhibitory interneurons (PV-In) showed increasing activity during SWS epochs, but unchanged activity during REM sleep epochs.
Our findings support the view that down-regulation of Pyr activity during SWS results from increased somatic inhibition via PV-In whereas down-regulation during REM sleep is achieved independently of such inhibitory activity. Overall, our findings show that SWS enables, through a spindle-related process, differential upregulation of cortical circuits likely involved in mnemonic processing, whereas REM sleep mediates general down-regulation, possibly through synaptic re-normalization.
P039 - Humans Can Predict Changes in Brain Activation. An Amygdala Electrical-Fingerprint Neurofeedback Study
1Department of Psychosomatic Medicine and Psychotherapy, Central Institute of Mental Health Mannheim, Medical Faculty Mannheim / Heidelberg University, Germany; 2Sagol Brain Institute, Wohl Institute for Advanced Imaging, Tel-Aviv Sourasky Medical Center and School of Psychological Sciences, Tel-Aviv University, Israel; 3Department of Physiology and Pharmacology, Sackler Faculty of Medicine, Tel-Aviv University, Israel
Nowadays it is well-established that individuals can learn voluntary self-regulation of brain activity through neurofeedback (NF). However, the understanding of underlying mechanisms is still limited. The question remains: Can humans be aware of their own brain states? Although leading NF learning theory posits that individuals can self-estimate their brain state, discrimination learning has been largely neglected by research. Moreover, it remains unanswered whether individuals can make metacognitive judgments about brain states, that is, whether individuals are conscious about the accuracy of self-estimation. In two empirical NF studies we investigated whether healthy participants could accurately estimate their brain activation. Using a within-subject repeated measures design, we administered up to 20 NF runs. We used the amygdala Electrical Fingerprint (amyg-EFP), a functional Magnetic Resonance Imaging-inspired Electroencephalogram surrogate of deep brain activation. Employing stepwise multilevel modelling, we explored on which sources of information participants base their self-estimation. As we were interested in internal monitoring of brain activation, we adjusted statistically for experience-based rating. Additionally, we investigated whether self-estimation accuracy predicts participants’ confidence about their rating. The analysis showed that amyg-EFP significantly covaried with ratings even after adjusting for experience-based strategies. This is in line with possible brain state discrimination. Unexpectedly, participants did not significantly improve in this skill. Higher rating accuracy predicted higher subjective confidence in the rating, which points to possible metacognition of brain states. This study provides initial evidence for awareness and metacognition in NF training with a marker from the affective brain circuit. The study was preregistered under https://osf.io/grxez.
P040 - Individual patterns of visual exploration predict the extent of fear generalization in humans
Generalization of fear is considered an important mechanism contributing to the etiology and maintenance of anxiety disorders. Although previous studies have identified the importance of stimulus discrimination for fear generalization, it is still unclear to what degree overt attention to relevant stimulus features might mediate its magnitude. To test the prediction that visual preferences for distinguishing stimulus aspects are associated with reduced fear generalization, we developed a set of facial stimuli that was meticulously manipulated such that pairs of faces could either be distinguished by looking into the eyes or into the region around mouth and nose, respectively. These pairs were then employed as CS+ and CS− in a differential fear conditioning paradigm followed by a generalization test with gradual morphs. Shock expectancy ratings indicated a moderately curved fear generalization gradient that is typical for healthy samples but its shape was altered depending on individual attentional deployment: Subjects who dwelled on the distinguishing facial features faster and for longer periods of time exhibited less fear generalization. Although both pupil size changes and heart rate responses also showed the expected generalization gradients, these responses were not significantly related to visual exploration. In total, the current results indicate that the extent of fear generalization depends on individual patterns of attentional deployment. Future studies evaluating the efficacy of perceptual trainings that aim to augment stimulus discriminability in order to reduce (over‑)generalization seem desirable. Data, materials, experimental & analysis scripts can be accessed via https://osf.io/4gz7f/.
P041 - Information load at learning determines the formation of schema memories at the expense of episodic details
1Institute of Medical Psychology and Behavioral Neurobiology, University of Tuebingen, Germany; 2Graduate Training Centre of Neuroscience, International Max Planck Research School, Tübingen, Germany; 3German Center for Diabetes Research (DZD), Institute for Diabetes Research and Metabolic Diseases of the Helmholtz Center Munich at the University of Tübingen (IDM), Germany; 4Center for Integrative Neuroscience, University of Tuebingen, Germany
The formation of semantic memories is assumed to result from the extraction of general, statistical knowledge across multiple experiences. At the same time, episodic details from the individual experiences might get lost during the abstraction of regularities among single events into schema-like memories. Thus, our investigation focused on the question under which circumstances schema-abstraction occurs at the expense of episodic details. For that, adult rats were subjected to a variation of the object place recognition task, in which animals were exposed multiple times to pairs of objects within an open field arena. Objects were positioned according to a spatial rule across learning trials. Learning either comprised a high information load, i.e. eight learning trials, or low information load, i.e. four learning trials. After learning, animals were left undisturbed for 24 hours, after which they were tested either for the expression of schema-memory, i.e. memory for the spatial rule, or episodic memory, i.e. memory for individual learning trials. We show that animals exhibit a more robust schema memory for the spatial rule in the high information load, compared with the low information load condition. In contrast, preliminary data suggests that animals only exhibit episodic memory for individual learning trials, when they were subjected to the low information load, but not the high information load condition. These findings suggest that whether individual memories are more likely to undergo abstraction into schema-memories or whether it is more likely that episodic details govern memory retrieval depends on the information load during initial learning.
P042 - Memory-based decisions depend on slower preference accumulation mechanisms than value-based decisions
1University of Basel, Switzerland; 2University of Hamburg, Germany
Decision neuroscience endorses the concept of preference accumulation in value-based choice formation. Accordingly, a decision maker integrates subjective value of different choice options over time until one option outweighs the other one by some margin. Although this mechanism applies to value-based decisions, it is unknown how preference is accumulated when subjective value needs to be constructed from episodic memory.
Over two sessions participants performed a remember-and-decide task, where they made choices between money offers and snack items. Snacks were presented visually (value-based) or needed to be recalled from memory (memory-based). In session 1 we obtained choices and response times (RT) and estimated a Drift Diffusion Model (DDM) which specified preference accumulation dynamics. In session 2 we measured EEG data to study the neural markers of preference accumulation.
While value- and memory-based choices were equally coherent with subjective value ratings, RT of memory-based decisions were significantly longer. Cognitive modeling indicated that longer RT in memory based-choices were due to lower drift-rate and higher boundary separation (in addition to increased non-decision time). Thus, the DDM predicted longer preference accumulation time in memory-based choices. This longer accumulation time was also reflected in EEG activity over central electrodes before motor responses, lending neural support for slower preference accumulation in memory-based choices.
Preference accumulation unfolds more slowly in memory-based choices but leads to similar choice consistency as value-based decisions. These altered accumulation dynamics may be related to hippocampal-cortical activity which is known to play a distinguished role in memory-based decision making.
P043 - Monetary incentives enhance intentional but not incidental subsequent recollection
1Universität zu Lübeck, Institut für Psychologie I; 2Universität zu Lübeck, Institut für Psychologie II
The positive effect of monetary reward on subsequent long-term memory performance is not consistent across studies. One possible explanation is that reward may have differential effects on incidental vs. intentional memory encoding. Here, we investigated this issue by using a monetary-incentive-delay (MID, incidental) and a monetary-incentive-encoding (MIE, intentional) task in 84 healthy human subjects (18-35 years old). In both tasks, participants had to respond as fast as possible to object images on day one and performed a modified remember-know recognition memory test approx. 24 h later. In the incidental MID task, participants received high vs. low rewards depending on reaction times on day one. In the intentional MIE task, participants also anticipated high vs. low rewards on day one, but received the rewards only for correct memory on day two. During encoding, participants responded faster to high vs. low reward images in the incidental MID task but not in the intentional MIE task. With regard to memory, high vs. low reward during encoding specifically promoted subsequent remember responses (i.e. recollection) in the intentional MIE task, but there was no effect of reward on memory in the MID task. Together, our findings suggest that reward promotes only intentional but not incidental subsequent hippocampus-dependent long-term recognition memory.
P044 - Noradrenergic stimulation reverses systems consolidation in humans
1Department of Cognitive Psychology, Institute of Psychology, Universität HamburgVon-Melle-Park 5, 20146 Hamburg, Germany; 2University Medical Centre Hamburg-Eppendorf, Department of Systems Neuroscience, Martinistraße 52, 20246, Hamburg, Germany; 3Department of Cognitive Neuroscience, Radboud university medical center, 6500 HB Nijmegen, The Netherlands; 4Donders Institute for Brain, Cognition and Behaviour, Radboud University, Kapittelweg 29, 6525 EN Nijmegen, The Netherlands
Introduction: It is commonly assumed that episodic memories undergo a time-dependent systems consolidation process, during which initially hippocampus-dependent memories eventually become reliant on neocortical areas. More recent research indicates that this process of systems consolidation is much more dynamic than initially thought; yet, whether the dynamics of systems consolidation can be shaped by environmental conditions, such as emotional arousal, remains unknown.
Methods: We probed the impact of post-encoding noradrenergic arousal on systems consolidation by combining a single pharmacological elevation of noradrenergic activity with fMRI scanning both during encoding and recognition testing either 1d or 28d later.
Results: Compared to placebo, intake of the α2-adrenoceptor antagonist yohimbine reduced the decline of memory over time. At the neural level, the placebo group showed the reorganization predicted by the systems consolidation theory: Hippocampal activity and multivariate encoding-retrieval pattern similarity, an indicator of episodic memory reinstatement, decreased from the 1d- to the 28d-delayed memory test, whereas inferior frontal gyrus (IFG) activity increased with time. This increase in IFG activity was directly correlated with the time-dependent memory decline. In sharp contrast to placebo, yohimbine led to a time-dependent increase in hippocampal activity and encoding-retrieval similarity from the 1d to 28d test, accompanied by a time-dependent decrease in IFG activity.
Discussion: These findings demonstrate that increased noradrenergic activity after encoding may not only decelerate, but even reverse systems consolidation in humans, leading to increased hippocampal involvement in memory and potentially more vivid memories over time.
P045 - Pharmacological increase of dopamine neurotransmission reduces decision thresholds during reinforcement learning.
Biologische Psychologie, Universität zu Köln
Introduction: Dopamine supports reinforcement learning by signaling expectancy violations. Pharmacological enhancement of dopamine transmission via l-dopa may improve reward learning relative to the D2 receptor antagonist haloperidol (Pessiglione et al., 2006). Here, we re-visited this finding in a larger within-subject study.
Methods: Healthy male participants (n=31) completed a reinforcement learing task (gain condition from Pessiglione et al., 2006) during fMRI directly after a previously reported four-armed restless bandit task (Chakroun et al., eLife, 2020). Participants received the dopamine precursor l-dopa (150mg), the D2 receptor antagonist haloperidol (2mg) or placebo (counterbalanced within-subjects). Data were modeled using reinforcement learning drift diffusion models (Pedersen et al., 2017) via hierarchical Bayesian estimation using JAGS.
Results: There was little evidence for drug effects on accuracy, total winnings or median response times (all BF01 > 5). Model comparison and posterior predictive checks revealed that in each drug condition, the data were best accounted for by a drift diffusion model with separate learning rates for positive and negative prediction errors. Boundary separation was reduced under both drugs (directional BF > 50, 95% HDI beyond ROPE).
Discussion: We did not confirm improved learning from positive reinforcement following l-dopa vs. haloperidol (Pessiglione et al., 2006). Instead, our data suggest similar effects of both drugs on boundary separation, compatible with a presynaptic autoreceptor-mediated enhancement of DA transmission via a low dosage of haloperidol (Wagner et al., 2020).
P046 - Reliving emotional memories: Episodic recollection re-elicits affective psychophysiological responses
1University of Amsterdam, Niederlande; 2University of Heidelberg, Germany
Introduction: Episodic recollection allows us to vividly and holistically re-experience past events. Affective psychophysiological states that are associated with past events may provide crucial information for the adaptive value of episodic memories and should therefore be re-instated when remembering past events. We tested whether episodic recollection was holistic to the extent that affective psychophysiological states are re-expressed.
Methods: In two experiments, participants encoded positive, negative, and neutral movie clips (N1 = 48, N2 = 68). One day later, they remembered the movie clips in response to neutral screenshots from the movies. On both days, we measured facial electromyography of the zygomaticus major and the corrugator supercilii that quantify the psychophysiological expression of positive and negative affect, respectively. The second experiment comprised a preregistered replication and extension of the first experiment (tinyurl.com/PUG-Duken). Data and code will be made available upon publication.
Results: In both experiments, encoding and remembering positive events elicited zygomaticus responses, while encoding and remembering negative events elicited corrugator responses. However, the intensity of affective psychophysiological responses during encoding did not consistently predict the intensity of the responses during remembering.
Discussion: Episodic recollection led to a replicable psychophysiological expression of affect but there was no consistent association between responses during encoding and remembering. Affective psychophysiological responses to memories may result from constructive imagery processes rather than a mere replay of past emotions. Given the adaptive value of affective states and their pronounced effects on episodic memories, episodic memory research should incorporate affective psychophysiological responses in theories and experiments.
P047 - Sleep-independent generalization of visual texture discrimination skills
Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, 72076 Tübingen, Germany
Introduction: Sleep-dependent consolidation of visual experience has been shown to improve texture discrimination. This improvement is thought to be specific for the trained location in the visual field. In the current study, we tested if previously acquired visual skills can be generalized to different locations and if such generalization benefits from sleep.
Methods: We trained healthy young participants on a visual texture discrimination task in only one visual quadrant before a night of sleep or daytime wakefulness. After the retention interval, participants were tested the following morning (Sleep group) or in the evening of the same day (Wake group) in both the trained as well as an untrained quadrant. Participants were explicitly informed about this testing procedure.
Results: Visual texture discrimination was better after retention compared to training, as suggested by reduced threshold stimulus onset asynchronies (SOA). Importantly, performance was not different between trained and untrained visual quadrants. There was also no difference between Sleep and Wake groups.
Discussion: Our results suggest that, unlike previously assumed, visual texture discrimination skills trained in a specific location can generalize to different locations in the visual field. Surprisingly, this effect appears to be independent of sleep, possibly related to prior explicit knowledge of both trained and untrained quadrants being tested after the retention interval. Future studies should determine the conditions in which visual skill knowledge does and does not generalize, with a particular focus on explicit vs. implicit instructions and the role of sleep in this context.
P048 - Transformation of visual memories in children and adolescents
1Department of Neuropsychology, Institute of Cognitive Neuroscience, Faculty of Psychology, Ruhr University Bochum; 2Life-Span Cognitive Neuroscience program, Institute of Gerontology, Wayne State University, Detroit, MI, USA; 3Helen Wills Neuroscience Institute, University of California, Berkeley, Berkeley, CA, USA; 4Pediatrics and Neurology, Wayne State University, Detroit, MI, USA; 5Neurodiagnostics, Children Hospital of Michigan, Detroit, MI, USA; 6Department of Psychology, Wayne State University, Detroit, MI, USA
The abundant sensory inputs that we receive across the day contain much more information than we need to remember. In fact, it has been argued that episodic memory does not only rely on the precise reinstatement of information but also on representational transformations that reflect the extraction of relevant features and statistical regularities. How these memory-related representations change during development is largely unknown.
We investigated neural representations during a visual recognition memory task in 35 children and adolescents aged 6-21 years (16 female) who were implanted with intracranial EEG (iEEG) electrodes prior to epilepsy surgery. We applied representational similarity analysis on iEEG time-frequency features and compared item-specific encoding-recognition similarity of subsequently remembered vs. forgotten scenes. We further analyzed differences in the magnitude of reinstatement across age.
We found significantly higher reinstatement for remembered vs. forgotten scenes in visual association cortices, including late occipital cortex (V3-5), inferior temporal cortex, as well as in parahippocampal cortex. Interestingly, this pattern reversed in early visual area (V2), where remembered scenes were significantly less similar to the corresponding encoding patterns than forgotten scenes, indicating a functional benefit of memory transformation in this area. Crucially, the magnitude of this transformation increased with age.
Our findings demonstrate that exact reinstatement of memory traces is beneficial in higher order visual association cortices, but detrimental in early visual areas. The age dependency of this effect may explain developmental gains in memory, possibly reflecting more efficient encoding strategies or an increasing dependency on conceptual features of the stimuli during recognition.
P049 - Understanding the appetitive mechanisms underlying relief learning
1Erasmus University Rotterdam; 2Biological Psychology, Clinical Psychology, Psychotherapy
Startle response is an automatic defensive response, which is potentiated by threatening stimuli, but attenuated by safety stimuli. Strikingly, the termination of threatening events elicits appetitive responses (i.e., relief) and stimuli associated with relief also attenuate startle reflex. Hearth rate variability (HRV) seems to mediate the safety-related startle attenuation. Here, we wanted to extend previous findings and investigate the role of HRV on relief-dependent startle attenuation. Fifty-two participants learned that one stimulus (fearCS) was shortly presented before an electric shock (unconditioned stimulus, US), one stimulus (reliefCS) was presented shortly after the US, and one stimulus (safetyCS) was never associated with the US. During the following summation test phase, both fearCS and reliefCS as well as fearCS and safetyCS were presented in compound. We found successful fear acquisition for both verbal and physiological responses meaning that fearCS compared to both safetyCS and reliefCS was rated more aversive, elicited startle potentiation as well as larger skin conductance response (SCR). During summation test, both SCRs and startle responses were significantly attenuated by the compound safety/fear, while the compound relief/fear did not reduce startle potentiation. In summary, conditioned fear was inhibited by safety signal, but not by relief signal suggesting that relief-associated stimuli may entail distinct appetitive properties.
P050 - Where is the toaster? Interplay of episodic and semantic memory during remembering of past events
Ruhr Universität Bochum, Deutschland
The scenario construction is a crucial part of remembering. Cheng, Werning & Suddendorf (2016) proposes in their new framework that during scenario construction, only the gist of an event is remembered episodically while missing details are substituted by semantic information. In our study we tested this prediction behaviorally.
Therefore we used a desktop based virtual housing environment. We created a conflict between episodic and semantic memory by placing objects in other-than-expected rooms (e.g. a toaster in the bathroom) or congruently to their semantic category (a toaster in the kitchen). We further manipulated the salience of the objects by making an object taskrelevant or not. Object memory was measured with three tasks - free recall, recognition-based room-recall and a spatial recall. Participants conducted them one day and one week after encoding.
Overall, both congruence and taskrelevance predicted successful memory retrieval. We could distinguish between correct, episodic recall (a toaster was recalled in the bathroom) and semantic substitution (a toaster was recalled in the kitchen). Specifically taskirrelevant objects were equally likely sorted to the correct room and to the semantically fitting room on both recall sessions during cued recall measures.
Thus our results provide a behavioral basis for the scenario construction model. Participants actively reported to have encountered objects in the semantically fitting room rather than in the correct room while remembering an episode. In a next step, we want to further illustrate this finding by investigating the neural basis of semantic substitution using fMRI.
P051 - Working memory performances do not seem to correlate with attentional behavior for iconic gestures in speech comprehension: preliminary results
University of Mons, Belgium
Iconic gestures (IG) convey meaning semantically related (i.e., congruent) to the speech they accompany. Studies on visual attention (VA) showed that listeners fixated mainly the speaker’s face and minimally IG. However, listeners appear to benefit from the presence of IG, particularly when presented with degraded speech. Recent studies have suggested an involvement of verbal and visuospatial working memory (WM) in the sensitivity to gesture-speech integration. The present study explores whether verbal and/or visuospatial WM performance could explain attentional allocation to IG in clear and degraded speech. One hundred and twenty-eight healthy French-speaking participants (35 men; Mage = 21.34; SD = 0.21) took part in the study. They first completed two WM tasks; the Digit Span Task (DST) and the Block Tapping Test (BTT). Then, fitted with an eye-tracking device, they performed a computerized task where they were simply asked to observe videos of an actor uttering short sentences and performing an IG. Regression analysis showed that in the presence of congruent IG and clear speech, performances at the BTT explained 3,9% of dwelling time on IG (R² = .039, F(1,126) = 5.11; p = .02). No other result was significant. While the presence of an association between visuospatial performances and VA to IG could be consistent with previous authors suggesting the creation, by IG, of a visuospatial context affecting language processing, globally, these results suggest an absence of relation between verbal/visuospatial performance and VA to IG. A subsequent study could investigate whether executive WM capacity could predict VA to IG.