Memory, Attention, and Perception (MAP) Lab
In the MAP lab, we investigate the cognitive neuroscience of visual memory and visual attention, capitalizing on the well-known functional–anatomic organization of visual perception. We employ multiple techniques, including quantitative psychology (modeling the mind), computational neuroscience (modeling the brain), behavioral studies, EEG/ERPs, fMRI, TMS, and research with patients to shed light on the complex nature of processing in the human mind and brain (see primary lines of research, below). All Ph.D. students in the MAP lab complete the three-paper dissertation option (see graduate-student first-author publications figure, below). Our research often runs counter to the majority view, with a focus on novel cognitive/neural processes and debated topics (see debated topics, below), which makes the MAP lab an interesting place to work!
Primary Lines of Research
Signal detection models of memory. We derive mathematical models of item memory and context memory based on signal detection theory, with target and lure distributions in decision space, to accurately characterize the nature of memory. Different models (e.g., the unequal variance, UEV, model, the two-high threshold, 2HT, model) are fit to empirical data to determine which model provides the best (and an adequate) fit. When existing models cannot account for the data, we develop new models (e.g., our source misattribution, UEVSM, model, shown in green below).
"Remember" spatial memory zROC (circles) with best fit models. Only the UEVSM model provided an adequate fit indicating memory is a continuous process (Slotnick, 2010).
EEG/ERP studies of memory and attention. We employ EEG phase–frequency analysis to investigate brain region interactions (e.g., phase-locked activity in the gamma or alpha bands) during item memory and context memory. ERP components are also evaluated to provide insight into the brain mechanisms mediating memory and attention.
Power spectra of thalamic and scalp electrodes during object recall (in successive 1 s epochs). These results indicate visual memory retrieval is mediated by widespread alpha decreases (reflecting disinhibition) followed by spatially-specific phase-locked gamma increases (reflecting binding; arrows indicate frequency bands of interest; Slotnick et al., 2002).
Computational models of fMRI data. We model fMRI data based on key parameters (i.e., acquisition volume dimensions, noise distribution, spatial autocorrelation, original and resampled voxel dimensions, individual-voxel type I error rate). Our MATLAB script (cluster_threshold_beta.m) runs Monte Carlo simulations to estimate the minimum cluster size required to correct for multiple comparisons. This method has advantages over others (e.g., no assumptions are made regarding the shape of the cluster) and is widely used in the fMRI community. We routinely improve the script in an effort to more accurately model physiological data.
fMRI physiological null data (left) and simulated null data (right). These results show that null data is uniformly distributed across the brain and can be well modeled using computational methods (Slotnick, 2017).
fMRI studies of memory. We employ fMRI to identify the brain regions associated with item memory and context memory (typically in conjunction with ERPs and TMS). In recent years, we have gone beyond the standard general linear model analysis and focused on more advanced methods such as functional connectivity analysis, multi-voxel pattern analysis (MVPA), and MVPA prediction analysis.
Spatial memory ROI MVPA correlation matrix. Correlations between patterns of activity for participants of the same sex were greater than for participants of the opposite sex, which illustrates sex differences (Spets, Jeye, & Slotnick, 2019).
Graduate-Student First-Author Publications
Debated Topics
Is recollection a threshold (all–or–none) or a continuous (graded) process?
The minority view, championed by Andrew Yonelinas, is that recollection is a threshold process. For a representative publication, see Yonelinas (1999).
The majority/our view is that recollection is a continuous process. We have found that the threshold recollection model never adequately fits the data (only sometimes appearing to fit due to artifactual flattening of the ROC when noise is included in the analysis), while a continuous recollection model (either the standard unequal variance signal detection model or a variant of this model) routinely fits the data. Our view is supported by the following publications:
Can spatial attention rapidly modulate activity in primary visual cortex (V1)?
The minority view, championed by Steven Hillyard, is that it can not. For a representative publication, see Martinez et al. (1999).
The majority view is that it can. The minority view is based on null findings, and there have been numerous instances of rapid attention modulation of V1, such that the null hypothesis can be rejected. Our view supported by the following publications:
Are standard methods to correct for multiple comparisons in fMRI analysis valid?
The minority view, championed by Thomas Nichols, is that standard methods are not valid. For a representative publication, see Anders et al. (2016).
The majority/our view is that standard methods are valid. We routinely update our cluster_threshold_beta.m script, which is widely used in the fMRI community to correct for multiple comparisons. Our view is supported by the following publications:
Are there sex differences in the brain?
The majority view, championed by Daphna Joel, is that there aren't sex differences. For a representative publication, see Joel et al. (2015).
The minority/our view is that there are sex differences, which is supported by the following publications:
Slotnick, S. D. (Ed.). (2021). Sex differences in the brain. Cognitive Neuroscience, Issues 3 & 4.
Spets, D. S., Fritch, H. A., & Slotnick, S. D. (2021). Sex differences in hippocampal connectivity during spatial long-term memory. Hippocampus, 31, 669–676.
Slotnick, S. D. (2021). Sex differences in the brain. Cognitive Neuroscience, 3–4, 103–105.
Is the hippocampus associated with only explicit long-term memory?
The minority view, championed by Charan Ranganath (for working memory) and Katharina Henke (for implicit memory) is that the hippocampus can be associated with working memory and implicit memory. For representative publications, see Roberts et al. (2018) and Schneider et al. (2021).
The majority/our view is that the hippocampus is associated with only explicit long-term memory, which is supported by the following publications:
Steinkrauss, A. C., & Slotnick, S. D. (in revision). Is implicit memory associated with the hippocampus?
Are all TMS protocols safe?
The majority view, championed by Joel Voss, is that all TMS protocols (currently employed by cognitive neuroscientists) are safe. For a representative publication, see Wang et al. (2014).
The minority/our view is that there are some TMS protocols, such as the 20-Hz multi-day TMS protocol, that appear to have long-lasting effects (beyond the participant's time in the laboratory) and other TMS protocols (e.g., continuous theta-burst stimulation, cTBS) that need further testing before being deemed non-invasive, which is supported by the following publication:
Is the hippocampus associated with only context memory (that requires item-context binding)?
The majority view, championed by Charan Ranganath, is that the perirhinal/entorhinal cortex processes item information (during item memory), the pararhippocampal cortex processes context information, and the hippocampus binds item and context information (during context memory). For a representative publication, see Diana et al. (2007).
The minority view is that the hippocampus can be associated with item memory or context memory (see Squire et al., 2007), with this region proposed to process multiple attributes of an event (Wixted & Squire, 2011). This view was more recently echoed by György Buzsáki who believes the hippocampus serves a general binding function, writing "the same general computation is performed on all incoming signals to the hippocampus, largely irrespective of their neocortical source. Whether a particular experimental observation implies that the hippocampus is computing space, time, memory, planning, abstract concepts, or other relationships may depend on the experimental design and the cortical input rather than on hippocampal computation per se” (p. 858, Buzsáki & Tingley, 2018).
Our view is that both sides of the debate have merit. For instance, we agree with the majority view that the hippocampus is typically associated with context memory to a greater degree than item memory, as context memory is usually associated with a relatively greater amount of information binding. However, we agree with the minority view that item memory can also be associated with the hippocampus (e.g., when binding multiple features of an item).
Our views are supported by the following publications:
Spets, D. S., Fritch, H. A., & Slotnick, S. D. (2021). Sex differences in hippocampal connectivity during spatial long-term memory. Hippocampus, 31, 669–676.
Is the fusiform face area (FFA) a face processing module?
The majority view, championed by Nancy Kanwisher, is that the FFA is a face-processing module. For a representative publication, see Kanwisher (2010).
The minority/our view is that the FFA is not a module and face processing is distributed across many cortical regions, like other objects. Although the view that the FFA is a face-processing module is still in the majority, the debate has been largely settled by the growing body of evidence indicating face processing (like all object processing) is mediated by a distributed brain network (i.e., face processing is not restricted to the FFA, and the FFA processes other objects than faces). Our view is illustrated in the following publications:
Can visual mental imagery activate primary visual cortex (V1)?
The minority view, championed by Zenon Pylyshyn, is that it can not. For a representative publication, see Pylyshyn (2002).
The majority/our view is that it can. This debate is largely settled as there is a large body of evidence indicating that V1 is associated with and necessary for visual imagery. Our view is supported by the following publications:
Are the dorsolateral prefrontal cortex and ventrolateral prefrontal cortex associated with spatial working memory and object working memory, respectively?
The minority view, championed by Bradley Postle, is that they are not. For a representative publication, see Postle et al. (2000).
The majority/our view is that they are. This debate is largely settled as the minority view was based on null findings, and there have been many significant findings indicating distinct lateral prefrontal regions associated with spatial and object working memory (i.e., one white crow would be sufficient to counter the null hypothesis, and there are lots of white crows). Our view is supported by the following publications: