Ben Kanter @beneuroscience@qoto.org

"When we hear a great talk, we recognize how we may be able to use the conclusions (and whether we want to), integrating them into how we think about, well, everything pertaining to the subject at hand. The same conclusions, offered during a bad or mediocre version of the talk, may well lose this potential (unless the listener works very hard to ignore the distractions of the presentation to reach the conclusions from the data alone). When we leave a really great seminar, we can understand and communicate those conclusions."

https://journals.biologists.com/jcs/article/136/7/jcs261147/306217/We-need-to-talk

Following Elsevier's decision to raise the article processing charge for NeuroImage to $3,450, all editors (inc. chief editors) from NeuroImage and NeuroImage:Reports have resigned, effective immediately.

I am joining this action and have also resigned.

Full announcement: https://imaging-neuroscience.org/Announcement.pdf

Mootstrapping: Running a bootstrap analysis to prove that two obviously dissimilar distributions are not in fact the same (p < .05)

On explanations in brain research:

A thread of the same idea comes up again and again in brain research. It's the notion that identifying the biological details (such as the brain areas/circuits or neurotransmitters) associated with some brain function (like seeing or fear or memory) is not a complete explanation of how the brain gives rise to that function (even if you can demonstrate the links are causal). To paraphrase:

Mountcastle: Where is not how https://www.hup.harvard.edu/catalog.php?isbn=9780674661882
Marr: How is not what or why http://mechanism.ucsd.edu/teaching/f18/David_Marr_Vision_A_Computational_Investigation_into_the_Human_Representation_and_Processing_of_Visual_Information.chapter1.pdf
@MatteoCarandini: Links from circuits to behavior are a "bridge too far" https://www.nature.com/articles/nn.3043
Krakauer et al: Describing that is not understanding how https://www.cell.com/neuron/pdf/S0896-6273(16)31040-6.pdf
Poppel: Understanding brain maps does not formulate "what about" the brain gives rise to "what about" behavior https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3498052/

Any other explicit references to add to this list? @Iris, @knutson_brain, Anyone?

Also, I imagine that some form of the opposite idea must also be percolating: the notion that 'algorithmic' descriptions of the type used to build AI will be insufficient to do things like treat brain dysfunction (where we arguably need to know more about the biology to, e.g., create drugs). Any explicit references of that idea? @albertcardona @schoppik, @cyrilpedia, Anyone?

"Not even wrong"

The physicist Wolfgang Pauli liked to use the phrase "not even wrong" to describe theories that failed to make testable predictions. You hear it now and again, and it's most often used disparagingly.

Theories can be "not even wrong" for different reasons. First, they are by their nature untestable (that's an unsolvable problem). Second, they are incomplete (a solvable problem and the first step toward a complete theory). The problem with the latter is when someone sells a "not even wrong" theory as fully baked.

To what degree do you see "not even wrong" theories as a problem in brain/mind (and other types) of research?

#neuroscience #cognition

HIPPOCAMPAL HISTORY TOUR PART 12: Charan Ranganath

#hippocampus #hippocampusHistory
#hippocampusGurus

Today's comments come from a contemporary of Mike Hasselmo's -- and someone I've enjoyed exchanging ideas with from the very first time we met. And it continues to this day. Charan's comments will have to be broken into two parts.

Charan Ranganath

What got you interested in the hippocampus?
I know it is a cliche, but what got me interested was hearing about Brenda Milner’s work with H.M., reading her work with temporal lobectomy patients… and also the general contrasts between declarative and nondeclarative memory described by Larry Squire, Neal Cohen, and Dan Schacter.

Really, though, all that work was part of my bigger interest in memory, and my first love was the prefrontal cortex. Having gotten my feet wet as an undergraduate in Art Shimamura’s lab, I was doing clinical neuropsychology in grad school. Later, having read Joaquin Fuster’s eloquent book on the prefrontal cortex and Marcia Johnson’s work on Source Monitoring, I was sold. That interest was reinforced by testing so many people in the clinic with frontal dysfunction.

I later fell into the hippocampus when I was doing my postdoc in Mark D’Esposito’s lab to study how the prefrontal cortex contributes to working memory and long-term memory. I started to read the monkey lesion and single-unit recording literature, and I noticed that many of the tests used to test the role of the prefrontal cortex in working memory were hardly different from the tests used to investigate effects of medial temporal lobe lesions on memory in monkeys. At the time, people who were doing standard tests of declarative memory in fMRI studies could not get hippocampal activation, and surprisingly, we found hippocampal activation when people were asked to do a working memory task. Our publication even included a replication–in the exact same voxels that we had investigated in the original study. Chantal Stern had gotten similar results with an n-back working memory paradigm. No one wanted to believe our results and I took a ton of flak from people–our paper was rejected by a number of top journals. But the findings were replicated by others, and other labs that ran our paradigm in amnesic patients found that hippocampal damage could impair working memory. I soon did a deeper dive, and I saw all these flaws in the previous literature on working memory and long-term memory, and I began to question all the standard dogma about the hippocampus. That led me to do all sorts of other studies to challenge the standard model of hippocampal function.

Aside from your own work, what findings about the hippocampus (and related brain parts) in the past 50 years most excited you, and why?
I think the most exciting and eye-opening thing was the discovery of what the hippocampus doesn’t seem to do, and that the “related brain parts” really do most of the heavy lifting in many memory tasks. One study that stands out was Betsy Murray’s study showing that the perirhinal cortex is the essential player for object recognition memory, not the hippocampus (as well as the work from Howard Eichenbaum, John Aggleton, etc.). Then I went back and noticed that Larry Squire’s lab had shown relatively minor memory deficits in humans with hippocampal damage (compared to those with perirhinal damage). Putting that together with the failure to find hippocampal activation in fMRI studies of “declarative memory” made me rethink all of my assumptions about what the hippocampus does.

Then there were the studies that revealed something different about hippocampal function. For instance, I was really influenced by Jen Ryan and Neal Cohen’s study, suggesting that the hippocampus may be critical for spatial relational memory even in the absence of awareness. I was also amazed by the convergence between Andy Yonelinas’ work in human amnesics and Howard Eichenbaum’s parallel studies in rats showing that the hippocampus seems to be critical for recollection. The Nadel and Moscovitch review on retrograde amnesia was another paper that took a while to pass through my defenses, eventually leading me to the realization that the standard systems consolidation theory actually made no sense. All of this stuff led me to really think about “context” and its central role in memory.

On a related note, context is also connected to a completely different area of research on the “P300” event-related potential, which is a neural response to unexpected events or targets in simple detection tasks. Donchin and Coles proposed that the P300 is an index of “context updating,” a process that we now call “event segmentation”. Bob Knight had shown that the P300 was abolished in patients with hippocampal damage, and many intracranial EEG studies showed that the P300 is ubiquitous in the human hippocampus. P300 effects are far more robust than any other neurophysiological signal in the human hippocampus. Thomas Grunwald basically demonstrated that if a hippocampus doesn’t generate a P300 it is non-functional.

Finally, I was drawn to the “time cell” studies from Gyuri Buzsaki and Howard Eichenbaum’s labs, and Howard’s work examining representation of item and context information in the hippocampus. All that stuff led me back to the foundational work on spatial memory done by John O’Keefe, Richard Morris, Edvard and May-Britt Moser, etc.

And since you mentioned “related brain parts,” the confluence of findings showing a critical role for the parahippocampal cortex (and possibly also the retrosplenial cortex) in all aspects of context representation in memory was probably the biggest bombshell for me, although the importance of that discovery has been overlooked by much of the field.

Can you relate one personal story about interactions with colleagues that most exemplifies the world of hippocampal research?
This is the question that held me up. It is hard to pick one because our community has a lot of different personalities. I can tell a couple of stories of social interactions that laid bare the behavior of highly influential scientists who abused their power in ways that really hurt vulnerable people both personally and professionally. But I can equally tell of stories of incredible senior scientists like Howard Eichenbaum, Ray Kesner, and you Lynn, who would talk to students, postdocs, and junior faculty members (male and female, white or POC) like what they had to say was incredibly important. That gives you a kind of boost that is hard to put into words.

I’ll choose one story which captures the most collaborative and constructive aspects of our world. In 1999 or 2000, at the Cognitive Neuroscience Society meeting, I literally bumped into Andy Yonelinas as he was presenting a poster on his work showing that hypoxia patients had impaired recollection and intact familiarity. At the time, I was a true believer in this whole thing that the hippocampus does “declarative memory,” and I told him that I was skeptical about this whole recollection/familiarity thing. It would have been reasonable for him to blow off this silly comment from a postdoc who knew almost nothing about cognitive theories of memory. Instead he totally disarmed me when he said, “You should be skeptical!” What followed was a cool conversation/debate about the topic and how the distinction between recollection and familiarity could be tested with fMRI. We designed a study together and made a bet that if the hippocampus and perirhinal cortex contributed differently to recollection and familiarity, I’d buy him a beer, and if not, he’d buy me one. For a while, it looked like I was right but when we got enough data to do a group analysis, it turned out he was right–and that result pretty much changed my view of the hippocampus, and in turn about what memory is in the first place. By now I owe Andy a lot of beers and probably will not live long enough to pay off my tab.

#hippocampus #hippocampusHistory #hippocampusGurus

HIPPOCAMPAL HISTORY TOUR PART 12: Charan Ranganath...part 2

What would tell a young researcher interested in the hippocampus to focus on now?
Think bigger–in terms of the brain, in terms of tasks, and in terms of models or theories.

There is a paradigm shift on the horizon. The idea that we can understand (at the computational or functional levels) the hippocampus as a module, independent of its connections, is untenable. Menno Witter’s detailed neuroanatomy, for instance, has made it very clear that the functions of the entorhinal cortex are inextricably linked with the hippocampus, such that you cannot understand one without the other. And the entorhinal cortex is the master hub of the brain, interacting with other major hubs, such as the prefrontal, perirhinal, and parahippocampal cortex, as well as subcortical areas like the amygdala. Then we can talk about neuromodulatory inputs like dopamine, cortisol, estrogen, etc. All this means that the computational environment of the hippocampus, by extension, its functions will be volatile and context dependent.

Moreover, the idea of the hippocampus as storing static “engrams” that are simply passed on to the cortex to reinstate memories is becoming increasingly untenable. We know that hippocampal representations are subject to substantial updating, but also they are dramatically modulated on the fly. A hippocampal “memory” of the same experience can be used for different purposes, and the representational space that you see in the hippocampus can look totally different depending on one’s goals or task context.

This brings us to the computational level. All the functions assigned to the hippocampus–which include everything from statistical learning and transitive inference to forming hyper-specific, pattern-separated memories–are computationally incompatible. They cannot be done by a single brain region. We can’t understand the hippocampus in isolation–we have to understand it in the context of dynamic interactions with other brain areas. That means we need to understand the characteristics of the animal’s environment (in a broad sense, not simply in terms of spatial geometry), its prior knowledge, and its goals. Trying to understand the hippocampus by asking people to memorize words in a word list, stretch out bird necks, or chasing fruit loops in an empty box is going to tell us nothing beyond what it does in those degenerate contexts.

If the goal is to understand episodic memory, we need to study events that have the complexity and timescale of events in the real world, and we need to consider how humans understand events in real time. If the goal is to understand navigation in a meaningful sense, we need to study navigation to a goal, in an animal with knowledge of its environment. If we want to understand the relationship between synaptic plasticity and forgetting, we need to study an animal that has more than two meaningful events in its lifetime, because retaining memories means maintaining robust connections amidst a daily barrage of interference.

So I think we need to study tasks that have a larger spatiotemporal scale, and think about the brain in terms of dynamic large-scale networks, and develop large-scale models that bridge these levels. This is going to be hard work and not for the faint of heart.

If you want me to get specific, one big opportunity that is waiting to be grabbed is to understand gating and regulation of information flow in the hippocampus. Right now, there’s a widespread implicit assumption that whenever hippocampal cells are active, that the neural activity is somehow behaviorally relevant. That assumption is probably wrong. For instance, we don’t think that the visual cortex is doing meaningful stuff when you sit around in the dark. Why should we assume that the hippocampus is constantly being consulted in the service of encoding and retrieval? There’s compelling evidence to suggest that, in fact, we don’t use the hippocampus constantly, and that a lot of what people see in hippocampal activity is just housekeeping. So, we need to understand when and how the hippocampus is brought in and out of action to serve behavior.

@tdverstynen @paulgribble

Agreed entirely. This is part of my grant writing advice.. sure you can squeeze in that extra sentence, but if you make the reviewer angry in the process, it had really better be worth it.

can we pleeeease stop producing documents (preprints, papers, grants, open reviews, ...) using full-justified sans-serif single-spaced text? We have decades and decades of design work on fonts and document styles that make it easier for human beings to read text, can we please take advantage of that?!

created by drawing with my eyes. I ALS and am a quadriplegic. This drawing is done using #eyegaze camera and ArtRage app.

#als #alswarrior #alsawareness

Came across this interesting paper on the limitations of permutation as a method for computing the p-value of a test: https://academic.oup.com/bioinformatics/article/22/18/2244/317881
#stats #pvalue #permutation

HIPPOCAMPAL HISTORY TOUR

#HippocampusGurus #hippocampusHistory #hippocampusGurus

The well has run dry. I am awaiting further promised reminiscences, and I can promise a very short publication lag.

Perhaps now would be a reasonable time to raise questions brought up by the personal histories published so far.

But, to contribute something historical today - I'm sometimes asked: why did you and O'Keefe write a book about your theory, since that's a pretty unusual choice for scientists, at least back it was back then. Our initial plan was to write a Psychological Review article, which is what one did with theoretical contributions in those days. But I ruled that out by unknowingly insulting the Editor of Psych Review at a sherry party in Elizabeth Warrington's office off Queen Sq. one evening in the early 1970s. I knowingly insulted a pompous speaker, but I didn't know he was the Psych Review Editor. A fact John O'Keefe conveyed to me shortly after we left the party. What started as a 50 page article aimed at Psych Review became a 350 page hand-typed draft that we circulated for comments to about 30 leaders in the field, including philosophers, psychologists and brain researchers. It took nearly 6 years to revise and publish the final version.

To Any Recently Laid-off Tech Workers,
If you have strong ML/Python skills, consider applying this position in my lab at MIT:
https://careers.peopleclick.com/careerscp/client_mit/external/jobDetails/jobDetail.html?jobPostId=25473&localeCode=en-us
The pay is lower and the free food not as fancy, but we do cool research on the human brain and have a lot of fun. And I have a near-perfect record helping my lab techs get into top Ph.D. programs in cog sci/ neuroscience.
Check us out here: https://web.mit.edu/bcs/nklab/index.shtml

Some thoughts on #BigBrainIdeas

Writing a book, I read a lot and I one thing jumps out at me: a continuum (from clarity to muddled) in how big ideas, ideas and facts are presented. I wonder if nearly every scientist would benefit from thinking a bit more in this way: What big idea am I pursuing? What are the more specific ideas under this umbrella (that remain bigger than hypotheses)? What's the evidence? Where are the holes? What parts of this space remain largely untouched (and why)?

A masterclass in laying out ideas is this 2001 Annual Review from Simoncelli and Olshausen on Efficient Coding:

https://www.cns.nyu.edu/pub/eero/simoncelli01-reprint.pdf

Reading it, you see that these two clearly understand what they are working on, why they are doing it, how it fits in with everything else that's happening in brain research, what the evidence is, and what parts of the space remain largely unexplored. Inspiring!

In comparison, there's a trend for reviews to launch with smaller ideas absent context, focus largely on the author's own work, and leave you with a disoriented sense that this was an ad campaign rather than a survey.

(BTW: the former also makes for the most brilliant type of faculty job talk whereas the latter fails to connect - they're most definitely related).