Yale Psychiatry Grand Rounds: October 15, 2021

October 15, 2021

Flynn Lecture: "Dynamic Reward Signaling in Ventral Pallidum"

Patricia Janak, PhD, Bloomberg Distinguished Professor, Department of Psychological and Brain Sciences, Department of Neuroscience, Johns Hopkins School of Medicine

Information

ID: 7045
To Cite: DCA Citation Guide

Download Transcript

00:00But we're still going to have
00:02people joining for awhile,
00:03but I I would like to make sure that I leave.
00:05There's not much time as possible
00:07for Doctor Janik to to speak,
00:09so I will begin our introduction.
00:12Those of you who heard The Chieftains,
00:15this music was in honor of
00:19Doctor Flynn's Irish heritage,
00:21although I do think The Chieftains
00:23might be Scottish, as Jane mentioned,
00:25but the IT was still lively and Celtic, and.
00:31As as you'll hear Doctor John Patrick Flynn,
00:36who for whom this lecture is name is named,
00:40was a member of the L faculty from 1954
00:43until his retirement in July 1979,
00:46and he was really an extraordinary
00:48person who had quite a remarkable life.
00:51And so I'm going to take a
00:52little time to tell you about it.
00:54We have his daughter,
00:56Sarah Flynn, with us here today.
00:58Thank you for coming, Sarah, and.
00:59She was the one who helped me
01:02gather the information that I'm
01:04going to be sharing with you today.
01:07So first of all,
01:08I I knew Dr.
01:09Finn Flynn's work because of his focus
01:12on the neural basis of aggressive
01:14behavior and he's recognized as
01:16a pioneer in neuroscience and in
01:18general for his contribution to
01:20understanding the function of the
01:23hippocampus in the hypothalamus.
01:25He also served from 1968 to 1978
01:28as director of the Abraham Ribicoff
01:30Research facilities at the Connecticut
01:32Mental Connecticut Mental Health Center,
01:34which is where most of our basic or
01:37a large chunk of our basic science
01:40labs remain right in in proximity
01:43to our clinical research facilities,
01:46which was something that was essential
01:49for establishing the translational
01:51and collaborative nature of the
01:53department and since 1982.
01:54We've had a lecture held in his
01:57honor recognizing his quote pivotal
01:59role in establishing the central
02:01importance of basic neuroscience
02:03research as their frontier for
02:05clinical psychiatric studies.
02:07And that's a tradition that we
02:09honor today with our speaker,
02:10doctor Patricia Janik.
02:12So here's now why we had Celtic music.
02:15Doctor Flynn was born in Superior, WI.
02:17The of an Irish immigrant mother and
02:20a first generation Irish American
02:22father who worked as a railroad
02:25switchman and he studied for the priesthood.
02:28He was ordained in Rome and then he
02:30returned to the United States in 1938
02:33to Loyola University and when his
02:35superiors there decided that they
02:37needed someone to teach psychology,
02:39he volunteered to study it.
02:41And then he went out.
02:43To find the best teacher so he
02:45could actually teach his students.
02:47This led him to Columbia University
02:49and there he studied psychology
02:51and he remained in the priesthood.
02:53But throughout this time he
02:55was examining his conscience.
02:57And he ultimately resigned from the
02:59priesthood and left the church in 1944,
03:02having received his PhD in
03:05experimental psychology in 1943.
03:07In 1944, Dr Flynn went to work at Harvard,
03:11where he did war work related
03:13to aviation and audition,
03:15and in late 1945 he married
03:18a holder Isma Garvey,
03:20and she was someone who I met
03:22when I first
03:23came to to Yale,
03:25and she would come with Sarah.
03:27To listen to the lecture and she was also
03:30an incredible and remarkable individual.
03:33She was a psychologist and in 1946
03:36and she was here actually as a member
03:39of the Department of Psychiatry,
03:42first appointed in 1962 as a research
03:45assistant and then serving on the
03:47planning project for the Connecticut
03:49Mental Health Center and finally.
03:52Working closely with Doctor Boris Astrachan,
03:55who was instrumental in founding the
03:58the Connecticut Mental Health Center,
04:00she she retired as a valued member
04:02of the medical school faculty.
04:05So in 1946,
04:08Doctor Flynn was appointed head
04:10of the psychology and Statistics
04:12Division at the Naval Medical
04:14Research Institute in Bethesda,
04:15and there he began his work
04:18in physiological psychology.
04:19And here's where the story gets even
04:22more interesting despite excellent
04:24performance reviews and general
04:26acclaim by his fellow scientists.
04:28Doctor Flynn was fired in 1953,
04:31and he was deemed a risk to
04:33national security for his quote.
04:35Close and continuing association
04:37End Quote with his wife,
04:40whose name had been named before the house.
04:42A committee on UN American activities
04:44during the McCarthy ERA era.
04:47Because an error is a correct
04:51Freudian slip because of of whole,
04:54this political activities in the 1930s
04:57and early 40s and doctor Flynn was
05:00offered the chance to keep his job if
05:03he divorced and he of course declined.
05:06Over the next six months or so,
05:08he received offers of employment from
05:10colleagues across the country at 13
05:12universities, and he told his daughter,
05:15Sarah Flynn,
05:15that each time his name reached
05:18the provost's office,
05:19the colleague was informed that
05:22the university could not hire Dr.
05:24Flynn, and in September 1954,
05:27Yale hired him to work with
05:30Doctor Paul McLean,
05:31who then held a joint appointment
05:33in Physiology and Psychiatry and
05:34who was studying the limbic system.
05:36So yell was able to benefit from
05:40his neuroscience area addition
05:43in the face of strong headwinds.
05:46Come upon a learning of so.
05:52Doctor Flynn then became a member
05:54of this of the department and
05:56worked until he retired in 79 on the
06:00physiological basis of aggression
06:03and he really made a a an incredible
06:07mark on the department,
06:09and upon learning of his death,
06:11Fritz Redlich,
06:12whose chair of the Department of Psychiatry,
06:13wrote to holder Flynn at John
06:15was all I ever wanted to be.
06:18A fine scientist and teacher,
06:20and most of all,
06:21an extraordinary human being.
06:22I've always admired his courage
06:24and integrity, two virtues,
06:26high value above anything else,
06:29and similar sentiments were
06:30expressed by other colleagues both
06:32at Yale and around the world.
06:33And the last thing I want to say before
06:36I I move on is also about Sarah Flint.
06:40Generate generosity to the department.
06:42So in April 2005,
06:45Sarah donated Dr.
06:47Flynn's most prized possession,
06:48the three volume set of romantica
06:51halls fixed Judah,
06:52their system and nervioso de Lumbre.
06:54Elizabeth brought us.
06:56In the original Spanish to the
06:59Yale Medical Historical Library
07:00and inside the first
07:02volume is an inscription written
07:04by Cahal in 1910, which reads in
07:08translation because of the brain.
07:10Man is the king of Creation,
07:12and to clarify the structure of the
07:14brain is to understand why that figure
07:16is at the head of the animal Kingdom
07:18and how civilization was created.
07:20A sign of human superiority
07:22to the rest of the beings.
07:23This may not actually translate
07:25so well to the current.
07:27Sarah, my original idea was that I would
07:29not be able to take a sure step in
07:31the study of Physiology and pathology
07:33of the nervous system without knowing
07:35the cerebral machine with precision,
07:37and that the mysteries of the science
07:39of the spirit will only be clarified
07:42when all the unknowns relative to
07:44the chemistry of the fine structures
07:46of the nerve cell are cleared up.
07:49And luckily, of course,
07:50we're completely done with that,
07:51and Doctor Janik will will give the
07:54the the heading to that to that quote.
07:58And just to finish up,
08:00Doctor Flynn always took delight in
08:02telling the story that he acquired
08:04these books for $10 at a used bookstore
08:07in New York sometime during the 1940s.
08:11So I I will finish there and I I hope
08:16that you will now join me in welcoming.
08:20Sorry I will get through.
08:22These are 2021 lecturer in the
08:26Flynn Memorial Lecture series,
08:28Doctor Patricia Janik and so Dr Janik
08:33is the A Bloomberg distinguished
08:35professor at Johns Hopkins University
08:37with appointments in the Department
08:39of Psychological and Brain Sciences.
08:41And the Krieger School of Arts
08:43and Sciences and the Department
08:45of Neuroscience in the School of
08:47Medicine and Doctor Janik Studies.
08:50Neural processes of reward learning.
08:52And you'll hear a lot about that today.
08:54She's especially interested in learning
08:56mechanisms underlying addiction,
08:57which is an area where this department
09:01certainly has extremely strong interest.
09:04She earned her pH.
09:05D.
09:05From the University of California,
09:06Berkeley,
09:07and then she conducted postdoctoral research
09:09at Wake Forest and at the National Institute.
09:11And drug abuse and come in from 1999 to 2014,
09:15which is the period when I first
09:18came to know her.
09:19She was faculty at the University
09:21of California at San Francisco,
09:22where she was the Howard J Weinberger,
09:25MD endowed Chair and addiction
09:27research at UCSF.
09:29She's a pioneer in the
09:31identification of neural circuits,
09:33underlying alcohol and drug seeking,
09:35and her work is really spanned
09:38levels of investigation from the
09:40molecular and synaptic plasticity.
09:42All the way to in vivo mechanisms
09:44that are relevant to complex models
09:45that are relevant to addiction,
09:47and this includes alcohol and
09:49drug seeking relapse,
09:51habit learning,
09:52extinction learning,
09:53and she's used.
09:54Everything from electrophysiological
09:56approaches to neuronal
09:57imaging and optogenetics,
09:59which I'm sure you'll hear about today,
10:01and if you're looking for a
10:02resource to understand the neural
10:04circuitry relevant to addiction,
10:06you need to read her 2021 review
10:08on consolidating the circuit
10:09model for addiction that she
10:11wrote with Christian luescher.
10:13And that appears in the annual
10:15review of neuroscience.
10:17And I first met Doctor Genich through her
10:18roles at the Society for Neuroscience.
10:20She's she's done a lot for the
10:22society she served as reviewing editor
10:24at the Journal of Neuroscience.
10:26She's been chair of the program committee,
10:28probably for way too long, given kovid,
10:30and she's incoming secretary of the society.
10:34She's also served on the Program Committee
10:37for the Research Society and Alcoholism,
10:39and She's been Co Chair and Chair
10:42of the Catecholamines and the
10:44Alcohol Gordon Conferences,
10:45so you can see that the
10:47influence of her work.
10:48In the field is extremely broad and
10:51I just want to close by saying that
10:53Doctor Janik is much more than her CV.
10:55She's been a mentor to leaders in the
10:57field who study the neurobiology of
10:59addiction and other behaviors that
11:01are relevant to psychiatric illness.
11:03When my own lab was trying to figure
11:06out issues related to experimental
11:08design for in vivo calcium imaging,
11:10Katie told me that the most important
11:12thing she ever learned in her scientific
11:14career was from Doctor Janik.
11:16And that is how to start with.
11:18Robust experimental design that gives
11:19you the adequate power to that's
11:22essential to get robust Physiology data,
11:24and we've certainly taken this
11:25to heart as well.
11:26So as you can tell,
11:28Dr Janik is a role model for many
11:31and she is the ideal person to give
11:34the Flynn Memorial lecture this year.
11:37So please let's welcome Dr.
11:39Janik and I will stop sharing and
11:42allow her to share her data with you.
11:47Thank you so much for that.
11:49It was such an an amazing introduction
11:52and I really enjoyed learning about.
11:56Now, Doctor Flynn and can
11:58you see my screen? Yes.
12:02I have to say
12:02what it, what a deep honor it is to be
12:04invited to give this particular lecture.
12:07I it was such an interesting history
12:09and I really hope that you find that
12:11the kind of work that I talk about
12:14today resonates with the kinds of
12:15things that he was interested in,
12:17so that you can see that it's a good fit.
12:19And I I want to thank you, Marina for
12:21the invitation and this opportunity.
12:23And thank you for such an A.
12:27Humbling introduction that that was really,
12:29really so nice and I'll try my
12:31best to live up to everything that
12:33has been said in this talk.
12:35So, uh, welcome to everybody and I'm
12:38sorry I'm not meeting you today in person,
12:41but I'm so happy to talk to you even
12:43though we're over zoom and if any
12:45questions come up during the talk.
12:47I'm sure people will help me to try to
12:49answer those since I don't think I'll
12:51be able to see the chat very well.
12:54OK, so I'd like to tell you about
12:57our experiments today.
12:58Looking at reward processing
13:00in the nervous system,
13:02specifically focusing on the
13:04area called the ventral pallidum,
13:07and I want to first tell you a
13:08little bit about the motivation
13:10for us in our lab in looking at
13:13reward seeking behavior models and
13:16the underlying neural circuitry and
13:18what we are interested in broadly,
13:20is what the processes are.
13:23The determined reward seeking behavior.
13:25Whether that reward is a food reward,
13:27something that our nervous system
13:29evolved to help us discover and ingest,
13:32or whether it's a drug reward,
13:34something that it's very important
13:35for us to understand as we think
13:37about how we can help individuals
13:39with substance use disorders,
13:41and we conceive of these processes
13:43through the lens of psychology.
13:46And there really are three interrelated
13:49psychological processes that
13:50determine at any one moment whether
13:52an agent will seek a given reward.
13:54And so first you have the.
13:55Real time decision.
13:56Will you decide to reach out
13:59your arm and grab that hamburger?
14:01Will the attic decide to call up
14:03the dealer to try to get that next
14:05fix so you have your real time
14:07decision that's impacted critically
14:09by your current motivational state?
14:12Whether you're hungry,
14:13whether you're thirsty,
14:14whether you are a person with an
14:16abuse disorder, whose craving drug,
14:18your decision is is necessarily filtered
14:21through your motivational state.
14:23And both of these critically depend
14:26on past experience or learning.
14:28So your past evaluation of the
14:31subjective effects of the rewards
14:33that you've experienced and you're
14:35learning about the conditions under
14:37which you obtain those rewards is
14:39critical for you in the future.
14:41When you're making that decision
14:43to get that reward so you know the
14:45actions to take or not take,
14:47and you understand the meaning of
14:49the stimuli in the environment.
14:51So we have these three interacting processes.
14:53And we're interested in the neural
14:55circuits that underlie them,
14:56so we can understand decision
14:59making both in normal
15:01conditions like feeding and also our
15:03end goal is to better understand this
15:06circuit so we can help explain decisions
15:09made by people who have substance use
15:12disorders or alcohol use disorders,
15:14because these same processes of
15:16course are occurring when one makes
15:19the decision to continue taking it.
15:21Drink for example, or to take another.
15:23Hit of that drug.
15:26So through many, many decades of work
15:29in a nonhuman animals and in humans.
15:33We've discovered as a field.
15:36A group of interconnected circuits
15:38that are called the canonical reward
15:40seeking circuit and of course.
15:42This circuit as many of you know
15:45overlaps extensively with the with the
15:47limbic system so something that doctor,
15:49Flynn would have been very well acquainted
15:51with and a circuit with in which he
15:54would have spent much of his time.
15:56In his research efforts.
15:58I'm going to focus on a subset of
16:01regions within this circuit here
16:03depicted in this cartoon schematic
16:05from the rodent brain where we
16:07see the nucleus incumbents.
16:08The most ventral aspect of this striatum,
16:11where we see its output,
16:13then one of its outputs,
16:15the ventral pallidum,
16:16which is analogous to globis pallidus
16:18in more dorsal striatal circuits and
16:21dopaminergic input to these regions
16:23from the VTA and other areas that
16:25we know and love like the amygdala.
16:27And so my lab is very interested in.
16:29How these reward circuits evaluate
16:33reward when it's being experienced?
16:36And then how that current evaluation
16:38can impact how the animals learn
16:41about what just happened so that it
16:44can impact their future behavior?
16:46So how are rewards processed in this circuit?
16:49So we're going to focus today specifically
16:51on trying to understand how the ventral
16:53pallidum contributes to that process,
16:55and it's much more interesting than
16:57perhaps we once thought few decades ago,
17:00the ventral pallidum historically
17:02has been considered somewhat of
17:04a way station or pass through.
17:06For information from this striedl region,
17:09the accompagnes,
17:10but instead increasingly,
17:11we're understanding that really important
17:14integrative processing is happening
17:16at the level of the ventral pallidum
17:19that impacts reward seeking behavior.
17:21So I'm going to focus in on some
17:23experiments conducted in the lab that
17:25I hope can can illuminate the function
17:27of the ventral pallidum for us,
17:29and when you wonder what a brain region does,
17:32of course one of the most traditional
17:34ways to look is to get rid of that brain.
17:36Region so decades ago,
17:38lesions of the ventral pallidum
17:40were shown to decrease intake of
17:42drugs of abuse in animal models.
17:45So decrease opiate self administration,
17:47for example.
17:49So that tells us this this area is very
17:51important for reward seeking behavior,
17:53but to figure out how it's very
17:55instructive to go in and use
17:58electrophysiology to record the neural
18:00activity during the behavior so you
18:02can see what the neurons care about,
18:04and so a number of labs have used this
18:06approach and I'd like to tell you about.
18:08Some of the data that has
18:10already emerged that set
18:11up our thinking for our experiments.
18:13So these are our data obtained from in vivo
18:18electrophysiological recordings in rats.
18:20So electrodes are implanted
18:21into the ventral pallidum,
18:22and we're measuring extracellularly.
18:24The spike activity of nearby neurons,
18:26and what Jocelyn,
18:28Richard working with Howard Fields found,
18:30is that neurons in the ventral pallidum
18:33care about cues that tell the animal
18:36it has an opportunity to get reward,
18:38and so here we see an average
18:40increase in activity when the animal.
18:42Here's a Q, but.
18:43The increase in activity is much bigger
18:45when the animals actually motivated
18:47to press a lever to get the reward,
18:49so the the larger signal is from trials
18:52when animals press the lever to get reward
18:54and the smaller signals when they fail to.
18:56So motivation already is coming into
18:58play and modulating the way the ventral
19:01pallidal neurons respond to cues.
19:04The valence of a reward also modulates
19:06responses in the ventral pallidum.
19:08So, as you might predict,
19:10given the data I told you about lesions,
19:12the ventral pallidum cares
19:14about the nature of the reward.
19:16So in this example from Kendall
19:17at all this this classic paper,
19:20we see examples from 1 neuron
19:22and its response to two QS.
19:24One is a Q that predicts something good,
19:26sucrose solution and one is a
19:28cue that predicts something the
19:30animal doesn't like a salty taste.
19:31We see the neuron has a big
19:33increase in firing to the.
19:35Q.
19:35Predicting something good
19:36and not to the salt Q.
19:39But if we use pharmacology to
19:41deplete the subjects of salt,
19:43so we'd make them desire salt and want salt.
19:46We see suddenly that this neuron
19:48now responds with an increase
19:50in activity to the salt Q.
19:52Reminiscent of the response to the sucrose Q.
19:54So this is a clue that the Q responses
19:57responses in the VP are sensitive to
20:00the valence of the reward the animal
20:02is going to get, so it cares about.
20:05Motivational state it cares
20:07about the valence of the reward.
20:09And interestingly,
20:10in this work from bullies Lab,
20:12it also cares about what's
20:14actually happening in real time.
20:16So here we see an example of
20:18data from a paper recording from
20:20neurons in the ventral pallidum
20:22in subjects that receive the queue
20:25that predicts sucrose reward.
20:26The neurons respond to the queue,
20:28and they respond to the reward.
20:30But if on some trials you omit reward
20:32that you see that there is a decrease.
20:35In firing by these neurons,
20:37this tells us that neurons have
20:39an expectation of the reward that
20:42they should receive after the Q.
20:44This is reminiscent of a negative
20:46reward prediction error signal that
20:48many of you may be familiar with from
20:50thinking about how dopamine neurons
20:52respond when expected reward fails to arrive.
20:55So together these give us important
20:58clues that neurons in the ventral
21:00pallidum respond to cues.
21:02They respond to reward,
21:03and they do so in interesting manners.
21:05They care about the motivational state.
21:07They care about the valence of the reward,
21:09how much the subject likes the reward,
21:11and they have some sort of
21:14expectation information.
21:15They care about what's
21:16happening in real time.
21:17If the reward arrives or not.
21:19So these and many other lovely
21:21studies set the stage for the kinds
21:24of questions that David Ottenheimer,
21:26a graduate student in the lab,
21:28wanted to ask when he wondered about
21:30the details of how the outcomes
21:33themselves are processed by the neurons.
21:36In ventral pallidum.
21:36So David was a graduate student in my lab.
21:39He's now a postdoc in the
21:41Steinmetz and Stuber Labs,
21:42and he was aided through all of
21:45this with by Doctor Joslin Richard,
21:47when she was a senior scientist in the lab.
21:50She was our resident ventral pallidum expert.
21:52Doctor Richard now runs her own
21:54lab at the University of Minnesota,
21:57so together they designed a
21:59series of studies to allow them
22:01to understand better how ventral
22:04pallidum neurons encode natural.
22:06Word outcomes and so that David was
22:09interested in doing these studies
22:12in the setting of multiple rewards.
22:15Because eventually we'd like to
22:17understand how agents make choices
22:19among rewards because we'd like to apply
22:22this in the future to drug addiction.
22:24How to agents choose drugs or other rewards.
22:28So David began with very simple.
22:31Experimental designs where he could
22:33look at the activity of ventral
22:35pallidal neurons when rats were
22:37receiving more than one reward.
22:39So in this very initial simple design,
22:41he implanted electrodes into
22:43the ventral pallidum of rats,
22:45recorded extracellular spike activity.
22:47These are waveforms of example
22:49neurons that he recorded,
22:51and here's again our cartoon
22:52of the ventral pallidum.
22:53So the electrode tips are residing in the VP,
22:56and he exposed subjects to
22:59two different rewards liquid.
23:02Sucrose and maltodextrin.
23:03These are both carbohydrates,
23:05calorically equivalent,
23:06and that we know rats like they will
23:09drink then avidly and he exposed them
23:12to these rewards in recording sessions,
23:15and the rewards were delivered randomly
23:17so the animal didn't know which
23:19reward was coming on any given trial.
23:22Specifically,
23:22he played the queue so white noise Q,
23:26and when the animal heard the cue,
23:27the animal knew it could go to the port
23:29when it put its snout in the reward.
23:32Port Reward was delivered,
23:33and then there's an Inter trial
23:35interval and that occurs again
23:37and so the the reward cannot be
23:39predicted by the queue.
23:40The animal has to actually wait till
23:43the reward squirt it out before
23:45it knows what it's getting so we
23:47can compare the neural response
23:48to these two rewards.
23:50There's an interesting feature
23:51about these two rewards and that is,
23:53although rats love both of them,
23:56if you give them a full bottle of
23:58one or the other on their homepage,
23:59they'll drink it all up.
24:01If you give them two bottles.
24:02One with sucrose,
24:03one with maltodextrin at the same time,
24:06most rats prefer the sucrose,
24:08and that's what I'm showing here
24:10in this behavioral data figure.
24:12This shows the preference subjects
24:14have for sucrose over
24:16maltodextrin when tested in the
24:17home cage when they just
24:19have big bottles of both,
24:20they'll drink more of the
24:22sucrose than multidex turn.
24:24However, in this behavioral session
24:25where we're giving A Q and then
24:28squirting out maltodextrin or sucrose
24:30and they have to drink it in order
24:32to have the next trial happen,
24:34we see that the licking behavior
24:36when they're consuming the different
24:39rewards is almost identical.
24:41So we're left with a nice,
24:42very simple behavioral model where their
24:44preference for the two rewards is different.
24:47We know that based on these
24:49long term drinking studies,
24:51but their motor behavior during this
24:53particular very simple task is very similar,
24:56so that gives us a nice way to
24:57see what the signal related to the
24:59preference might be when we're
25:01basically making sure that the motor
25:03behavior is not that different,
25:04because that could motor behavior
25:06could be an explanation for
25:08some differences that we see.
25:10So in the face.
25:11In this very simple behavior,
25:13what David found when he recorded from
25:15many neurons in the ventral pallidum,
25:17many individual neurons is that there's
25:19a big difference in the way the neuron
25:22signal which reward they received.
25:24So here I'm showing the average activity
25:27of 205 neurons that were sensitive to
25:30reward based on statistical analysis.
25:33If we divide the trials into those in
25:35which the animal receives sucrose and orange,
25:38or maltodextrin in this pink purple color.
25:41We see an average very large increase in
25:43activity when the animals are drinking.
25:45The sucrose 0 is the time that rewards
25:48delivered the first few seconds,
25:50first three to four seconds is when
25:52they're actually lapping it up.
25:53We see a much lower response by the
25:56population when maltodextrin is received.
25:59These heat maps here show you the
26:01activity of the individual neurons
26:02that make up these averages,
26:04so again we have the same time course
26:07and each row is the color coded map of
26:09the spike intensity from that neuron
26:12arranged by most intense to less intense.
26:14And you can see here by I many,
26:16many neurons are showing an
26:18increase at this exact time,
26:19and it's much less present and sometimes
26:22even more of a decrease for maltodextrin.
26:25So the populations in the ventral pallidum
26:27encode these two rewards differently.
26:29Although the drinking behavior similar the
26:31preference these subjects have is different,
26:34and that may be what is we're seeing here.
26:37Alternatively,
26:38you might propose will.
26:39Sucrose is a very important natural sugar.
26:42Maybe neurons in the brain are set
26:45up already to fire in a very specific
26:48way to sucrose as as it taste it.
26:51So it could be that these responses
26:53are fixed and that they really
26:55depend on the rewards.
26:57So David tried to think of a way to to
26:59examine that so to do that he repeated
27:02the same behavioral procedure but he
27:05swapped water for sucrose. So now.
27:07He's going to give the animals up.
27:10Interleaved sessions,
27:11when they receive sucrose or water after
27:14the queue and you could see their behavior.
27:17In fact, they're licking behavior
27:18is different from water because
27:20they're not water restricted.
27:22So they don't really want water very much
27:25and what he saw neurally is a switch in
27:29the way neurons encoded maltodextrin.
27:31So remember the activity for Maltodextrin
27:33was much lower than for sucrose when
27:36those were the two rewards being compared.
27:38But now when we are comparing maltodextrin
27:41and water as the animals taste each one,
27:43we see a relative increase in the
27:46response from maltodextrin and a
27:47decrease in the response for water.
27:49And you could see that very
27:50clearly in these heat maps.
27:52Here's a sucrose here's water,
27:55but most interesting focus on
27:57maltodextrin relatively low activity
27:59across the population now very.
28:01High activity across the population.
28:04So this readout is not fixed based
28:06on the the actual chemical nature of
28:09the taste it so instead it it seems
28:12as if perhaps it relates more to the
28:14animal's current preference for example.
28:17And we can see very exactly similar
28:20results if we run a behavioral
28:22session with all three liquid's
28:24randomly presented after the Q,
28:272 hour rats and we see a much higher
28:30average neural response to sucrose medium
28:33for maltodextrin and big decrease for water.
28:36So so there there's a ranking in
28:39the neural activity that fits what
28:41we might think of as the ranking of
28:44the animal subjective preference.
28:46So that's one way we could.
28:48We could wonder about what the
28:49signal means for the animal.
28:51Is this just a readout of the animal's
28:54current subjective preference?
28:55Another idea that David had when
28:57looking at this is that this signal
29:00also could be a readout of a
29:02difference from the animals,
29:04expectation of reward value.
29:06So each time the animal,
29:08here's the queue and is going
29:10to the port to get reward,
29:12it doesn't know which rewards coming,
29:14so it would have the same
29:17average reward expectation.
29:18And then the animal might receive a reward
29:21better than average worse than average,
29:24you know,
29:24just slightly better than average.
29:26So this might also map onto what
29:29you might predict you would see
29:31with an expectation signal.
29:33So we were very interested in trying
29:35to figure out how could we tell the
29:37difference between a signal that might
29:39tell us something about if the animals
29:42using it to to read out violations of
29:45expectations or alterations of expectation.
29:49Or is the animal just using the
29:50signal to read up?
29:52Yes,
29:52this is what I like best.
29:53This is what I like worse.
29:57And this is basically a repeat
29:59of what I have just said.
30:02So the the way in which David decided
30:04to tackle this was to collaborate
30:06with the lab of Jeremiah Cohen,
30:09also at Johns Hopkins University,
30:11and his then MD PhD student Bill Albari.
30:14And so Jeremiah and Bella had been
30:16using quantitative models to try
30:19to explain the activity of neurons
30:22and what what they care about.
30:24And so in this way you might use a
30:26quantitative model and try to fit
30:28the firing rate of given neurons
30:30to aspects of your model to try to.
30:32Understand what those neurons are encoding,
30:35and so this is what Belal and David
30:38together in collaboration did to
30:39ask if there was any impact of
30:42expectations on firing at the time
30:44the animals drinking the reward and
30:47the way they did this was to look
30:49at the to the canonical Rescorla
30:52Wagner model that that tells us how
30:56predictions are updated by experience.
30:59And so this is the reward prediction
31:02error framework that many of you
31:04are familiar with and in this
31:06framework the expected value and
31:08animal holds for upcoming reward.
31:11Is updated iteratively,
31:13so with every experience based on
31:16whether the rewarded receives mattress,
31:18that or is better, or is worse,
31:21and so that's where we have
31:22positive prediction error if it's
31:24better than expected, no change.
31:25If it's the same as expected,
31:27negative prediction error is what you
31:28get is worse than what you thought,
31:30and through the use of this canonical.
31:34Way of explaining what the
31:37activity of a neuron might be,
31:41we can compare that with a much simpler idea,
31:44which is that the readout at the time,
31:46the animals drinking that reward just
31:48reflects a difference in outcome.
31:49A binary difference,
31:51sucrose versus maltodextrin?
31:53Or is this bike activity that we see at the
31:56time of reward unrelated to either of these?
31:58Of course,
31:59we already have an idea that many of
32:00the neurons do care about rewards,
32:02so we don't expect that that that will be a.
32:05Huge contributor so you can take this
32:07bike activity of neurons through time.
32:09Trial by trial.
32:10Look at how the neuron responds to the
32:13reward and see if its activity matches
32:15just a real time difference in outcome.
32:18Or does it actually account for what
32:20the animal received a trial before
32:22trial before the trial before,
32:24as in a Rescorla Wagner model,
32:26and I wouldn't be saying this if
32:28we hadn't indeed found a group
32:31of neurons that does care about
32:33what reward the animal received.
32:35The trial before the trial that
32:37they're experiencing that reward.
32:38In other words,
32:39there's an impact of experience.
32:41So in about 20% of these neurons
32:43that fire at the time of reward
32:46showed an impact of expectation,
32:48another slightly more than 20%,
32:51were just encoding the current outcome.
32:53This ones better, that was worse,
32:55and that was relatively stable.
32:57And then there were neurons that didn't
32:59care about either of those things,
33:01and you can now look at the neural
33:03activity of these different.
33:05Classes we now divided up our neurons
33:07that respond to rewarding to these
33:09two classes and get a nice feel for
33:12what this actually might look like.
33:14So here are the neurons that were the reward
33:16prediction error neurons they cared about.
33:18What happened trial before
33:19divide it up just into.
33:21The simplest kind of way of
33:24thinking about this trials,
33:25in which animals get sucrose.
33:28After a trial when they got
33:30maltodextrin so better than expected,
33:32that's this very tall yellow peak.
33:34Trials when animals got
33:36sucrose after sucrose.
33:38Trials when animals got multi dextrin.
33:40After maltodextrin and trials when
33:42animals got maltodextrin after sucrose
33:44so much worse than they thought and you
33:47can see this big modulation of firing.
33:49That depends on what just happened.
33:51So that matches this.
33:53This quantitative assessment
33:54and the current outcome.
33:56Neurons show much less or no modulation
33:59around expectation and is quite flat
34:02for the unmodulated neurons obviously,
34:04and we can do this same analysis in
34:06R3 reward tasks and find the same.
34:08Outcome where there's a portion
34:10of neurons that encode some kind
34:13of expectation signal looks like
34:14a reward prediction error,
34:16and even more neurons do this
34:18when you have this larger dynamic
34:20range across the rewards,
34:21the animals experiencing,
34:22which is kind of interesting,
34:24we can again get an intuitive feel
34:27for how this maps onto neural activity
34:30by looking at subsets of the neurons.
34:33In this case,
34:34if we just look at the neurons that
34:36are seem to encode and expectation.
34:38Signal and categorize them based
34:41on the calculated error.
34:43Was it very positive?
34:44Was there no error signal on that trial?
34:47Was there a negative error signal on
34:49that trial and use eight categories for that?
34:52We can see this beautiful
34:54distribution of signals along the
34:57most positive reward prediction error,
34:59little error and negative error.
35:01So this gives us a sort of intuitive
35:04way to think about how firing
35:06happening at the time of reward
35:08can actually be telling us.
35:10Think about what the animal expects,
35:12not just what is this particular
35:15reward as far as identity.
35:18So this this was really exciting to
35:20us to find this kind of signal and BP.
35:23We're very used to thinking about
35:25these kinds of expectations.
35:26Signals in dopamine neurons,
35:27and we hadn't been expecting to
35:29see this kind of thing in ventral
35:31pallidum at all,
35:32which we thought would be more just
35:34a basic readout. This is good.
35:36This is bad in real time.
35:38So because we saw these signals that
35:41match what are teaching signals,
35:43signals that are updated over time
35:46and reflect what subjects expect,
35:48David wondered if he could find
35:50any way to see if the animal's
35:53behavior changed based on what the
35:55error signal was on a given trial.
35:59So we have to say that the procedure
36:01that he designed was not designed
36:03to see a lot of rich behavior
36:05was designed to have the animals
36:08behavior very similar.
36:09Each trial. Indeed, you saw that when
36:11we were looking at the licking rate,
36:13but David did still videotape the animal,
36:16so he went back and analyzed their behavior.
36:19This is what the Chamber looks like.
36:21A typical rat chamber,
36:22and where the animals can enter reward port.
36:25You know when the queues on and drink
36:27the reward and there's an interest.
36:29Inter trial interval you
36:30know they wander around.
36:31They might do. A little grooming,
36:33might hang out by the port
36:34waiting for the next month.
36:35Typical behavior that the rat behaviors
36:37among us are used to seeing what
36:40David found when he looked at the
36:42behavior in detail that on trials when
36:45animals had just received the reward,
36:48they liked better sucrose.
36:50Right after that,
36:51the animal tended to hang around the port,
36:53waiting presumably for more sucrose.
36:55If the animal had just received maltodextrin,
36:58they tended to wander off more.
37:00It's not a huge effect.
37:02We can look at this.
37:03This is sort of a way to map
37:05individual animal behavior.
37:06Looking at a cartoon of
37:07the square of the chamber,
37:08it's not a huge effect,
37:10but you see more color and more black.
37:12X is away from the port when
37:15it's post maltodextrin sucrose,
37:17and this is statistically significant.
37:20We see the same pattern when we
37:21have the three rewards together.
37:23After receiving water,
37:24they're more likely to be
37:26further from the port,
37:27so it's very simple measure of how
37:31their future behavior is impacted
37:33by the the the the validation
37:36of their expectation or getting
37:38something less than they expected.
37:41Simple small effect,
37:42but can we manipulate it by changing
37:45these neurons and how they fire?
37:48So this was the setting for an
37:50optogenetic experiment at David,
37:52then performed getting together
37:53with Tabatha Kim and Kurt Fraser to
37:55graduate students in the lab at that
37:57time had now moved on to Charles
37:59River and to post up with stuff on
38:01the mouth and so they simply asked
38:03if they express general adoption.
38:05Excited Tori option in neurons
38:07in the ventral pallidum and then
38:09make those neurons fire.
38:11By shining light on them at the time,
38:13the animals ingesting reward,
38:15can they impact the behavior
38:17after the animal got the reward?
38:19How?
38:19How much the animals likely to
38:21hang out by the port?
38:22That's so if you excite the neurons,
38:25you'd expect to see the animals
38:26hang close to the port right
38:27after you've excited the neurons.
38:29Alternatively,
38:29if you express an inhibitory option
38:32in ventral pallidum so that you
38:35can inhibit BP neuron firing at the
38:37time they're drinking the reward,
38:40you'd expect to make.
38:41The animal more likely to be away
38:44from the court after that sort
38:46of inhibition of BP,
38:49and so that's the kind of
38:50experiment that they designed.
38:51This is just a cartoon by lateral
38:54inhibition of BP neurons is going to
38:56occur at the time of reward delivery
38:59or unilateral excitation of DP.
39:01Neurons will occur at the time
39:03of reward delivery.
39:04So for this experiment the
39:06reward is the same in all trials,
39:08it's sucrose, but the optogenetic.
39:11Activation or inhibition will
39:12just occur on half of the trials,
39:15so that lets you see if your change
39:17in neural activity is affecting the
39:19animals behavior independent of the
39:21specific taste of the reward, etc.
39:23So you hold the those aspects of
39:25the reward constant and see if
39:27you're turning up or turning down
39:29of the neural activity impacts
39:30their behavior as you would expect,
39:32and that is exactly what they
39:34observed in this Chamber.
39:36That reward port is here on the right side,
39:38on trials in which BP was activated.
39:41The subject tends to hang out
39:43closer to the port,
39:44so a lower value here than on
39:47trials where the subject was not
39:49did not receive VP activation,
39:51so it's a within subject within
39:53session comparison and we see
39:55the opposite with inhibition.
39:57They tend to be further away if you
39:59inhibit while they're consuming the reward,
40:01so this these are reliable effects but
40:05granted relatively small in this procedure,
40:07not really used to.
40:09Think about how this impacts decisions,
40:12but because we saw the signal
40:14in this behavior,
40:15we wanted to see if we could find
40:17any evidence that an expectation
40:19reward prediction, like signal,
40:21impacted future behavior and this
40:23evidence was there and so that
40:25was really exciting to us and
40:27lead the stage for our continued
40:29experiment that I'll tell you about.
40:31Right now I'm going to take a quick
40:33interim summary and this also be a good time,
40:36if anyone.
40:38Wants me to clarify something that I've said.
40:42So I wanted to just say from these.
40:44So far we've learned that the signal
40:46in VP that responds to reward is
40:48sensitive to pass reward history
40:50and can provide a reward prediction
40:53error signal to update the animals.
40:55Expected value of reward.
40:58And So what we would like to know,
40:59of course, is, are these signals used?
41:02Do they interact with decision processes?
41:05Can they impact the actions animals make?
41:07'cause that's ultimately what we
41:09want to explain how our choices made,
41:11what's going on in the brain when
41:15the animal evaluates the options?
41:18Question yes please.
41:20That's really beautifully done.
41:22I was wondering if you've
41:24looked at what happens with.
41:26Obviously, this is a learning signal.
41:29What if it's a pharmacologic reward that
41:32is sensitive to to tolerance effects?
41:35Or you know any sort of reduction?
41:38What happens to these neurons and to behavior
41:41'cause ya'll animal moved away
41:42when they were not getting,
41:44you know, beautiful extinction.
41:45But that is not obviously what
41:47we see when when people are
41:49beginning not even addicted.
41:50Just beginning to come
41:53to start to really like and
41:55escalate their their use.
41:57Yeah, that's a great question.
41:59So what we have not done yet is
42:00is used Ivy drug, for example,
42:02or even alcohol in this exact model.
42:05And we we we need to do that
42:07because here we are beginning to
42:09define for ourselves what these
42:11neurons are doing to natural reward.
42:13And when natural reward choices
42:15are occurring.
42:15But the critical question is how are
42:18these processes altered when that reward?
42:21Is a drug reward that has pharmacological
42:23properties that are quite different
42:25and there is important data from,
42:27for example,
42:27Megan Creed and Christian looser,
42:29showing that drugs like cocaine change
42:33synaptic efficacy between, for example,
42:36the accompagnes and ventral pallidal neurons.
42:39So there are chronic effects of drugs
42:42on the way that these neurons should
42:44be activated and should be firing
42:47and and so finding that intersection
42:49and studying that was, is it?
42:51Ago.
42:55OK so I'm gonna go on and tell you about
42:59the next reward behavioral procedure
43:02where David extended this work to try
43:05to understand how these signals are.
43:07Even if these signals matter.
43:09As far as the choices the animals make,
43:11and so he again turns to the notion of.
43:14Reward choice, so you need more
43:16than one reward at the same time
43:18and to provide a setting where he
43:19could try to understand choice.
43:21He thought it would make the
43:23most sense to make the animals
43:25choices change through the session
43:27by changing their motivation.
43:29He was interested in understanding how
43:32motivational state impacts choice.
43:34And I was very interested in this
43:36because motivational state and how
43:38it might be relieved by rewards you
43:40choose is a nice analogy for eventually
43:43thinking about how drug craving may
43:45operate within the system and how
43:47taking drugs may reduce craving.
43:49And then what happens in the brain.
43:51So that's the Longview.
43:52But in the short view,
43:54David wanted to ask specifically
43:56about motivational state shifts
43:58and how they may impact animals
44:01decision making through this system.
44:04Where we're recording expected
44:06reward and reward preference,
44:08and so he's doing this in the
44:10face of a shift in thirst.
44:12So obviously whether you choose
44:14food or water will depend on
44:15if you're hungry or thirsty,
44:17so this is a a kind of motivational
44:19shift that has great relevance to
44:20the natural functioning of this
44:22circuit and that we thought would
44:24help us understand the natural
44:26functioning of this circuit.
44:27So David developed what he called
44:30the dynamic preference task.
44:31So this task is very simple.
44:34Some animals are choosing between sucrose
44:37and water reward by pressing a lever.
44:41They begin each day,
44:42thirsty and within the session
44:45they assuaged their thirst.
44:47And within this session,
44:49besides the choice trials,
44:51it's critical that David also had
44:53forced choice trials where throughout
44:56time the animals had to experience
44:59water and experienced sucrose
45:00so that he could monitor the VP
45:03signals to that through the session,
45:05and so this will become clear when I
45:08explain again how this procedure works.
45:10So animals rats are in the
45:13behavioral chamber, their electrodes,
45:15and their ventral pallidum 60.
45:17Percent of the trials they receive
45:19over an hour and a half are the
45:21same as what we talked about before.
45:22There's a queue that tells them go
45:25to the reward port and 50% of the
45:27time they get sucrose. 50% water.
45:28It's randomized.
45:29They don't know what it will be.
45:31These are the forced choice trials
45:33they have to complete this to go
45:35on to the next trial.
45:3740% of the time they hear a cue that
45:39tells them it's a choice trial.
45:41They get to pick if they get.
45:42If they receive sucrose or water
45:45by pressing the relevant lover.
45:47So we have a mix.
45:48Of these outcome choice trials where
45:50we can see their behavioral preference,
45:52what do they want at that moment in time?
45:55And we also have the forest
45:57trials where we can't
45:58see their preference from their behavior,
46:00but instead we can look at how their
46:03neurons respond to the two rewards and
46:05see if it changes as their choices change.
46:07So we use both of these kinds of
46:10trials to get important behavioral and
46:13neural data that we want to relate.
46:15And what you see behaviourally when
46:17you look at the responses of a rat
46:19in this kind of procedure is that
46:22they start out choosing water.
46:23That's the long purple lines.
46:25This is session time and the
46:27number of trials which makes sense.
46:29They're thirsty, they're going to
46:30press on the water level quite a bit,
46:31and as they get less thirsty,
46:33they'll press on the water level less.
46:35They'll press on the sucrose level
46:37lever a few times in the beginning,
46:40but that increases overtime
46:42as they become less thirsty,
46:44so there's a shift in their choices,
46:48and you can graph that with this green line,
46:50which shows their relative preference.
46:52The short black lines tell us
46:54when the forced trials occurred,
46:56so you see they're forced to sample sucrose
46:58and water throughout the whole session,
47:01and we see there be choice behavior
47:03through these choice trials.
47:05So in all of the subjects.
47:06Used in this study that I'll talk about.
47:08We see a similar shift in preference
47:12through the behavioral session,
47:14so as they become less thirsty,
47:16they tend to just respond for the supers,
47:18which makes sense.
47:20So we see this behavioral shift.
47:23What about the neurons in the
47:25VP and so to to address this,
47:28David looked again at this response
47:31to reward that VP neurons emit,
47:34so he's looking at this time
47:36period just after reward delivery.
47:38When many neurons fire spikes when they
47:41get reward and he's using now a general
47:45little mini linear excuse me model to
47:48try to understand which aspect of of.
47:51Uhm,
47:52the the design best captures how neurons
47:55fire through session time trial by trial.
47:59Do they just tend to show a difference?
48:01Reflective of the difference in outcome?
48:03Sucrose versus water that's
48:05relatively stable over time.
48:06Do they just show a decrement
48:08or increase in activity?
48:10They start satiety as you move through time.
48:12Or is there an interaction
48:14between these two processes?
48:16And by looking at this statistically,
48:18David founded sizeable proportion
48:19of neurons that care about.
48:22Both of these at the same time,
48:23so their activity fits best.
48:25Changing preference through time,
48:27so some something to do with satiety.
48:31Presumably something to do with preference.
48:34And that makes sense.
48:36'cause that's what happens to
48:37the behavior with the behavior
48:38switches as you move through time.
48:40The animals preference for
48:42water versus sucrose switches as
48:44they become less thirsty,
48:46and so here on the left,
48:47if you can see this might be hard,
48:50but I'll describe it for you is just
48:53an example to show 1 neuron firing
48:56in a very typical way for the whole
48:59population through the session.
49:01So at the beginning we have
49:03neuron spiking at session.
49:05The first trials and at the end
49:07the last trials and these shaded
49:09areas are the times when the animals
49:11drinking sucrose or drinking water.
49:13These are the times analyzed and you can
49:16see that as the animal first gets sucrose,
49:19you see moderate spiking.
49:21That increases overtime when
49:22the animal first gets water,
49:25you see a lot of spiking.
49:26That really decreases overtime.
49:28If you look at this same kind of
49:32feature overtime for all of the neurons.
49:35Plotted here in these two figures,
49:37with the sessions divided into quarters,
49:39quarter 1234,
49:40you see that the mean reward response to
49:44sucrose is moderate and then gets bigger.
49:48You see that the mean response to
49:50water starts big and positive and
49:52gets smaller and more negative
49:54as the session goes on.
49:56So you can see this much more easily
49:58if we think about the mean here
50:01over quarters for sucrose versus
50:03water in this final graph down here,
50:05the bend firing rate,
50:07we can see the increase in activity for
50:10sucrose overtime in one session and
50:13the decrease in mean activity for water.
50:17So this is interesting because we
50:19see that the neural activity sort
50:20of shifts more excited for water
50:22in the beginning,
50:23more excited for sucrose at the end.
50:26And so does the animals preference, right?
50:28The preference of the animus shift similarly.
50:31But another thing to note is
50:33this isn't like a mirror image.
50:35These two curves are exactly symmetrical.
50:38This water line really decreases,
50:40and the sucrose mine is kind of flat,
50:42so this was pretty interesting
50:44and I just thought, well,
50:46that's the way the data are,
50:47but given David's beautiful more sort
50:50of quantitative mind, he thought,
50:52well, what?
50:53How can I explain that particular shape?
50:55Is there a way I can characterize?
50:57That quantitatively and he started
50:59thinking about whether this reward signal
51:02that signaling a reward prediction error.
51:05Could it contain more information
51:07than just something related to what
51:09reward did I get on the last trial?
51:11Could it also reflect the value of the
51:14whole task as the animals becoming sated?
51:18So,
51:18so every every both rewards will
51:20become less valuable in some sense,
51:22as the animals becoming less thirsty.
51:26So I I really thought this was a
51:28beautiful insight that he had and he
51:31developed based on his prior work with Bill.
51:33All models again to fit to the
51:36activity of each neuron to see which
51:39kind of quantitative model best
51:41explained the way the neurons fired
51:43and on the left you see the firing
51:46rate in a session of an example
51:48neuron just to remind us it's the
51:50increase in response to sucrose at
51:52at the top that orangey red line
51:54is moderate and there's a sharp.
51:57More dramatic decrease in responding
51:59to water in this blueish purple
52:01line and so you can ask if just
52:04a simple straight line satiety.
52:05Does that explain best the
52:08way the firing changes?
52:09Is the firing explained best
52:11by a preference switch that
52:13it would be perfectly symmetrical
52:15preferences just from you know zero to 1?
52:18Or what about both of these together?
52:21And so for David,
52:23that's just a linear combination.
52:25Of models describing both of these,
52:27and then when you combine these literally,
52:30it looks like this and you know you can
52:32see where I'm going because already the
52:34model shape looks similar to the neural
52:36shapes that we've been looking at.
52:38And so here again, is the model.
52:39When he looked at each neuron and fit its
52:42activity to these three different models,
52:44he finds that the best fit
52:47model is this mixed model.
52:49What this means is most neurons
52:51seem to care about both satiety,
52:53so movement through the session in time
52:56and their current preference for reward,
52:59which one they're liking better.
53:02So that's pretty cool.
53:04So, So what David showed us is
53:06that it's not just the immediate
53:08difference in reward value that is
53:10being reflected in this activity,
53:12but there's also an impact of satiety.
53:16So this was all analyzed based
53:19on forced trial data, right?
53:21'cause we're looking at how the animal
53:24responds to the reward through session time.
53:26But we have all of these choice
53:28trials for the animals making its
53:30own decision about which reward
53:32it wants at that given time,
53:35and what David wondered is,
53:37do these responses of the neurons
53:39that tell us how much the animal,
53:42what the animal thinks about the
53:45reward relative to its expectation,
53:47does that have anything to do
53:49with their behavior?
53:50Because in the end,
53:51we'd like to try to get an understanding
53:53of how these systems impact choice.
53:56That's our eventual goal.
53:57So the way that David decided
54:00to think about that.
54:02Was to look at the animals behavior.
54:05Good idea to do first.
54:06Here are three rat examples on the left.
54:09You're looking at the animals choice
54:11behavior through the session.
54:13The purple bars on the bottom are
54:15when the animal chooses water
54:16they do a lot at the beginning,
54:18last moving forward,
54:19and then they shift and tend to choose
54:21sucrose more as session time goes on.
54:23So you can plot the preference curve.
54:25For sucrose,
54:26the preference curve for water
54:28and you see this.
54:29This kind of function and you can see
54:32that example for three different rats.
54:34When David looked at the neural
54:37estimates for this mixed model that
54:41came from what he calculated here,
54:44it came from these forced trials
54:45and tried on a trial by trial
54:48to estimate what the animal's
54:50preference was just based on the
54:52neural activity for a given neuron.
54:54Then averaging that across
54:55all neurons from a given rat,
54:58you get these preference curves,
54:59and they're just remarkably similar.
55:02You don't even need all the beautiful
55:04statistics to tell you the neurons
55:06are giving the same readout of
55:08what the animals preferences,
55:10moment by moment as the decision
55:12the animal makes.
55:13So this is a really nice correlation
55:16that helps us build on the idea
55:19that this signal is important for
55:21the animals future decision, and.
55:23That's what these graphs
55:25here on the right support.
55:27If you look at the correlation for each
55:29neuron of its activity with the animals
55:32preference for neurons like that.
55:34Were weighted within this mixed model
55:36that care about outcome in time.
55:38We see that correlation is very
55:40close to one for very many of them.
55:42If you ask from the neural
55:45activity when in time this switch
55:47point might be for each neuron.
55:50Many neurons that care about outcome
55:53by time give you a very close
55:56estimation of the actual trial.
55:58So the switch point is zero.
55:59You can see many or within 20 trials.
56:03So, quantitatively,
56:03we've got a really nice agreement
56:06between neural activity with the
56:09animal actually decides to do.
56:11So that led save it again to turn
56:13to optogenetics to see if he
56:16could manipulate the system and
56:18manipulate the subjects choice.
56:19And this is the final little bit of data,
56:22bit of data that I'll be showing you and
56:24then we can discuss this as you wish.
56:27So now David wants to see if by
56:29controlling ventral pallidal neuron
56:31activity he can impact their choice,
56:34so so he has optimal control.
56:37He's now not going to make the
56:40animals thirsty, he's just going to.
56:41Go back to the situation where they're
56:43choosing between sucrose and maltodextrin.
56:46This is a situation when they're not
56:48thirsty and their behavior through
56:49the session tends to be relatively
56:51stable and he wants the behavior
56:54to be relatively stable because now
56:56he must to go in and try to change
56:58it by messing with the BP reward
57:00prediction error signal.
57:01So what he's going to do is express
57:04channelrhodopsin in ventral palatal
57:06neurons and shine light on them
57:08to force them to fire every time
57:11the animal drinks maltodextrin.
57:13And in the procedure that he uses,
57:15there's going to be forced choice
57:17trials so he can continue to make
57:20them drink maltodextrin paired with
57:22stimulation and choice trials so we
57:25know what the animal actually would
57:27prefer to drink at any given time.
57:30First. Animals are well trained.
57:32Then there is a session of
57:35optogenetic manipulation,
57:36so well trained animals.
57:37And this is just a diagram of when
57:40the stimulation occurs.
57:41So at the time that the animals actually
57:43drinking maltodextrin and we can talk
57:45about that more if you want to later.
57:47So if you look at baseline here on the right,
57:50a well trained animals prefer sucrose
57:52based on their choice trials.
57:55The press the lever to get sucrose
57:56most of the time and the blue dots are
57:58the subjects in which were expressing
58:00channelrhodopsin the Gray dots or
58:02subjects expressing the empty vector GFP.
58:04Both animals get laser shining in
58:07their brain, but the great author.
58:09Our control.
58:10In the test session,
58:11the next day after baseline were
58:13stimulating every time the subject gets
58:16less preferred reward maltodextrin and
58:18what you see is it shifts preference
58:21towards maltodextrin on the choice trials.
58:23So what does that actually look
58:26like through time?
58:27Here is one session, the test session.
58:30And here's the smoothed preference
58:32based on liver choice for each rat
58:35Gray or the controls relatively stable.
58:37They mostly want sucrose and blue.
58:40Are these experimental animals where we've
58:42shifted the preference to maltodextrin,
58:44and you can see it's a gradual effect
58:47that accrues through experience,
58:49so it's not that the first
58:50time you stimulate it,
58:51they immediately shift.
58:52This is congruent with the idea.
58:54But it's a learning response.
58:56You're sending them a signal that
58:57that reward is better than expected.
58:59So maybe you should change what
59:01you do on upcoming trials.
59:03And also congruent with the idea
59:05that this is a learning signal.
59:07This behavior change does last
59:10until the next day.
59:12So the test day is the day of optogenetic
59:16manipulation Recovery day one.
59:18We just see what their choices are
59:19and you can see that their choices
59:21still tend to be more towards
59:23maltodextrins then they weren't
59:25baseline and this changes over time
59:27as we no longer ascending that
59:29fake signal that we could send
59:31with the optogenetic manipulation.
59:33So this is a first step to providing
59:36some evidence that this signal can
59:39impact the animals future choice.
59:41So if you look at their.
59:43Latency's to choose levers in the
59:46optogenetic stimulation experiment,
59:48you find that overtime they tend
59:51to choose maltodextrin more,
59:53and on trials choice trials when
59:55they're going to choose maltodextrin,
59:56their latency to go to the lever
59:58to make that choice is faster,
01:00:00so we see this change in behavior
01:00:03that matches what you would expect.
01:00:05For this. This kind of signal.
01:00:08So what I told you is that initially
01:00:10we see a signal in the ventral
01:00:12pallidum when animals are actually
01:00:14experiencing reward ingesting it.
01:00:16That seems to match their relative
01:00:19preference at that current time,
01:00:21and if you analyze that spike activity
01:00:23relative to the current time and
01:00:25the reward period just before that,
01:00:28and just before that,
01:00:29and just before that,
01:00:30IE reward history,
01:00:31you see that at least a subset of these
01:00:34care about reward history and what
01:00:36they instead are signaling is a reward.
01:00:38Prediction error is what I just
01:00:40got better than I expected.
01:00:42The same or worse.
01:00:44And so these same signals,
01:00:47these moment by moment reward
01:00:50prediction error signals also care
01:00:52about the current motivational state.
01:00:55They're also able to integrate the
01:00:57larger change in value that might
01:01:00happen as your motivational state is
01:01:02changing as you get less thirsty.
01:01:04And hypothetically,
01:01:05in other situations we haven't tried yet,
01:01:07right?
01:01:08As your craving might be reduced
01:01:10as you ingest drugs.
01:01:12As hunger changes as you eat etc and
01:01:15so these signals that occur at the
01:01:18time the animals ingesting reward,
01:01:20they affect their future behavior.
01:01:22As we saw in that very simple.
01:01:25A measure of how close you are to the port,
01:01:28and as we saw in this,
01:01:29perhaps more informative choice
01:01:31procedure where animals are choosing
01:01:33which lever to push in order
01:01:36to get the reward that
01:01:37they want at that time.
01:01:39And so the big question,
01:01:40of course, is what?
01:01:42What do these signals mean
01:01:44for the circuit as a whole?
01:01:46So if I go back to the statement that
01:01:48I made at the beginning that usually
01:01:50the ventral pallidum was the more
01:01:52boring area that was just the output.
01:01:54For the fantastically interesting
01:01:56nucleus incumbents, of course,
01:01:58the nucleus of Cummins is
01:01:59fantastically interesting.
01:02:00But these are big excitatory signals
01:02:02in the ventral pallidum unlikely to
01:02:04be driven by the Gabaergic medium.
01:02:06Spiny neurons of the nucleus accumbens,
01:02:08and when we look.
01:02:09And David didn't record in
01:02:10the comments as well.
01:02:12When we look there,
01:02:13we don't see the large numbers of
01:02:15neurons representing this reward
01:02:16prediction error in the same way,
01:02:18so it's a it's a signal can built
01:02:21here in the VP most likely by
01:02:24integrating various inputs important
01:02:26new work from Megan Creed's lab at
01:02:28Saint Louis in Saint Louis showed
01:02:31that projections from the ventral
01:02:33pallidum back to the nucleus.
01:02:35Incumbents in fact,
01:02:36might be really important when animals
01:02:38are making decisions to consume.
01:02:41Rewards so,
01:02:42so the VP has a really interesting
01:02:44relationship with the rest
01:02:46of the reward circuitry,
01:02:47their inputs to to BTI and VTA projects,
01:02:50to ventral pallidum,
01:02:51and so,
01:02:52so how these signals that we
01:02:54identified fit in with the rest
01:02:56of the activity of the reward
01:02:58circuit is a really important.
01:03:00Future direction as well as trying
01:03:02to map how the circuit response
01:03:05to natural reward with how it
01:03:07might respond to drug reward.
01:03:09Because, as mentioned,
01:03:10your eventual goal is to try to
01:03:13understand these interactive
01:03:14processes and how they modulate
01:03:17in humans are seeking of things.
01:03:19We should seek our food rewards and
01:03:22think and seeking of rewards that in
01:03:25some individuals can become unhealthy.
01:03:28So I think I'll I'll stop there
01:03:31with just thinking again.
01:03:32The lab members that participated
01:03:35in this work,
01:03:37and I identified David and Jocelyn early.
01:03:40They're really the main drivers of this.
01:03:42I also showed Jude members of
01:03:44Jeremiah coincide that were important,
01:03:46but I want to thank the lab in
01:03:48general for all of their input for
01:03:50this work and lab meetings and
01:03:52and helping one another conduct
01:03:54all of the experiments that I
01:03:56want to thank funding from NIH,
01:03:57of course,
01:03:58and I want to.
01:03:59Thank you all very much for giving
01:04:01me the opportunity to talk about
01:04:03this basic neuroscience research
01:04:05and I hope it gives us all some
01:04:07ideas about how we can think about
01:04:09basic nice neuroscience work and
01:04:11how it can tell us about the human
01:04:14condition and how we doing basic
01:04:16neuroscience work can learn and shape
01:04:18what we do based on the human condition.
01:04:21So thanks very much.
01:04:24Thank you so much, I'm really
01:04:27enjoyed that and I have encouraged
01:04:29the trainees to ask questions first
01:04:32if there's any trainees out there.
01:04:35Doctor Taylor is a trainee of
01:04:37course lifelong but may not qualify,
01:04:40so I would like to start with with a trainee,
01:04:43but if not we can we can get
01:04:46to questions from UN trainees.
01:04:51Doctor Taylor's training
01:04:55all right? Well then that seems appropriate.
01:04:57Doctor Taylor.
01:04:57Why don't you kick us off?
01:04:59I have questions too. Sorry, I come.
01:05:03If a trainee wants to interrupt me,
01:05:06please go ahead.
01:05:08That was a beautiful talk as always Patricia.
01:05:12So I have a question which is
01:05:16sort of how dynamic do you think
01:05:20these VP responses are in that.
01:05:24I wonder whether you would see
01:05:28similar VP signals related to
01:05:31expectation and prediction error.
01:05:33If you in your experiment initially
01:05:36looked at sucrose preference
01:05:39compared to something that was
01:05:41actually slightly aversive,
01:05:43like a salt solution,
01:05:46but then made the animals
01:05:49physiologically salt induced the salt,
01:05:53a salt state for that reward.
01:05:56Would you see the VP neurons suddenly
01:06:00switch over to tracking the?
01:06:03The salt, uhm?
01:06:06Yeah, yeah, I I think so and so.
01:06:09So although this is a history
01:06:12dependent signal that follows this
01:06:14reward prediction error kind of model,
01:06:16we think it's not model free but it's
01:06:19more model based in that the subject
01:06:21can update it on the fly and so so you
01:06:26know in in this procedure animals start
01:06:29each day thirsty and then become sated.
01:06:32Their response to in a different
01:06:34procedure where there's a water
01:06:36predictive cue their response to the
01:06:38water predictive cue is very high
01:06:40at the beginning of each session,
01:06:42even though it's very low by the
01:06:44end of the session when thirst is
01:06:47is no longer a drive.
01:06:48So so there.
01:06:49I'm not sure if I'm getting to exactly
01:06:51what what your question was, but it's a.
01:06:53It's a super dynamic system and
01:06:55I think immediately impacted
01:06:57by the animals expectations,
01:06:59and it doesn't necessarily have to accrue.
01:07:03Overtime it's impacted by
01:07:04what happens over time,
01:07:06but also can be directed by a more
01:07:08sort of cognitive, a goal directed.
01:07:12Uh, evaluation.
01:07:16List Europe.
01:07:19Thank you for that amazing talk
01:07:20and I'm curious 'cause we've
01:07:22been talking about expectation,
01:07:23but most of the signal that
01:07:25you've been presenting is
01:07:26during the reward consumption.
01:07:28So do you see anything when the
01:07:30queue is present that indicates the
01:07:31outcome that they're expecting before
01:07:33they can compute the prediction
01:07:34error? Yeah, that's a great question.
01:07:37So in this particular set of
01:07:39studies that I told you about
01:07:41the queues were not informative.
01:07:42As far as the identity of the reward,
01:07:45so we didn't find very
01:07:47much expectations signal.
01:07:48In those cues,
01:07:50but David ran a variation in this
01:07:52dynamic thirst task where there
01:07:55were accused that came before
01:07:57each of the rewards in the forest
01:07:59trials of a water Q&A sucrose Q,
01:08:02and in that case the reward prediction
01:08:04error like signaling transferred
01:08:06to the queue as you would predict,
01:08:09and so that expectation related
01:08:11response was mostly there in
01:08:13the in the way that the animal
01:08:16responded to the queue.
01:08:17So that's that's a great question.
01:08:20Al
01:08:22hi, thanks so much for the talk.
01:08:24I was really struck by your result
01:08:26that about heterogeneity in terms
01:08:28of value versus RPE signals in
01:08:31the ventral pallidum, especially
01:08:33given this kind of ongoing
01:08:36discussion about whether a Cummins
01:08:38concentration of dopamine represents
01:08:39a value or RPE signal,
01:08:41and in particular. Some
01:08:45recent work of Sam Gershman,
01:08:46and now, which is showing that
01:08:47you can maybe reconcile these
01:08:49approaches by having a kind of RP
01:08:52signal with sensory feedback that
01:08:54looks like a value signal.
01:08:56And so I guess I was just served as an
01:08:58open ended question.
01:08:59Wondering whether you think that this
01:09:01heterogeneity in the ventral pallidum
01:09:03might somehow either resolve this
01:09:05discrepancy or correspond to as well
01:09:07the concentration of dopamine dynamics.
01:09:10Yeah, that's a great question and
01:09:12my short answer is is I I don't
01:09:15know so it is true that this system
01:09:17is highly interconnected with the
01:09:20canonical dopamine neurons in the VTA.
01:09:23So VP neurons project back to the VTA,
01:09:25both to dopamine neurons but also to
01:09:28the GABA interneurons and the VTA.
01:09:30Dopamine neurons do send a projection to VP,
01:09:33so there's some interaction in the
01:09:35creation of these kinds of dopamine
01:09:37signals that that we think about as far
01:09:39as our PE and. What exactly are they?
01:09:42Are they communicating to the incumbents?
01:09:44Uh, so I.
01:09:45I don't know how that's all gonna
01:09:47workout as far as trying to see if
01:09:49there's a separable value signal or
01:09:50whether it's really going to be able
01:09:53to be understood all as a readout of
01:09:56online changes in what the animal
01:09:59is actually actually receiving.
01:10:01So I think that's something that's
01:10:03still left to work out.
01:10:04So really interesting sort of not a problem,
01:10:07and the communication between the
01:10:09VP and the incumbents.
01:10:11Is also going to be a factor.
01:10:12Presumably so, uhm.
01:10:14I like the direction of your question
01:10:18because it forces these results have
01:10:22forced me and all this forces us to
01:10:25not think of reward prediction error
01:10:26is just here in this group of dopamine
01:10:29neurons and then medium spiny neurons.
01:10:31Maybe all are transmitting expected value
01:10:33and this area does this in this area.
01:10:35Does that and instead you know
01:10:37these systems are all interconnected
01:10:39and these variables seem to be
01:10:42represented to greater or lesser
01:10:43degrees throughout the circuit.
01:10:45So beautiful work from the U2 lab,
01:10:48in fact,
01:10:49has shown has shown that really
01:10:51painstaking recording studies,
01:10:53so it's it's going to be harder
01:10:56to figure out than then.
01:10:58I would dream when we want to make a
01:11:01nice model where everything is only
01:11:03just very very separable and in a
01:11:05separate kind of neuron and separate place.
01:11:08But that's a nice challenge.
01:11:09There's work to do it for the future.
01:11:13Ralph, I'm going to ask a quick question
01:11:15from the chat and then over to you.
01:11:17Media Naseer says rats prefer water
01:11:20over sucrose when dehydrated.
01:11:22However, some people prefer sugary
01:11:24soft drinks over water when thirsty.
01:11:26Is there a mechanistic
01:11:28difference in this situation?
01:11:30That's a great question, and one I I
01:11:32don't know the answer to and I would.
01:11:34I guess I would immediately wonder
01:11:37about the role of long term experience
01:11:40in humans for going for the sugary
01:11:43soft drink to relieve thirst and
01:11:46and whether we could model that.
01:11:49By the way we expose our rats
01:11:51to these different rewards.
01:11:53Overtime probably someone working
01:11:55more in the nutrition field
01:11:57maybe has already done that,
01:11:59so that answer might be known
01:12:00and I just don't know it.
01:12:01I think that's a great question.
01:12:04Ralph, go ahead.
01:12:06Trisha really like these
01:12:08experiments and great data.
01:12:09I have a question
01:12:11I guess about the maltodextrin and and what.
01:12:15Is this is this purely sensory?
01:12:17Is it post ingestive?
01:12:18I know the answer is probably both,
01:12:19but I guess I'm interested
01:12:21because maltodextrins are
01:12:22sort of a tricky thing.
01:12:23It's not very sweet,
01:12:24but of course you can't detect
01:12:25it at high concentrations.
01:12:27I can't remember
01:12:28from from David early paper how
01:12:30high it was, and if you know
01:12:32that Jamaican detector at least
01:12:34taste it. And of course some
01:12:35of the signals you're seeing
01:12:37or so fast is clearly something
01:12:38century. But I wonder if over the session
01:12:40they're learning. It has a really
01:12:42it gets broken down so quickly into glucose.
01:12:44It's guys seeing indexes.
01:12:45Is higher than sucrose, right?
01:12:47So it should be really fast post
01:12:48ingestive signals and so I wonder
01:12:50how you how you think about that.
01:12:53I think, uh, so David more than
01:12:56me really did the important work.
01:12:59All you train is out there of
01:13:00looking at all the old literature,
01:13:02including all the old animal
01:13:04behavior literature where people
01:13:06have done a lot of work comparing
01:13:08tastants and looking at preference.
01:13:10Ways to measure preference in assessed
01:13:12preference in rodents and so.
01:13:13So he he looked carefully at that when
01:13:16he chose maltodextrin as a comparison.
01:13:18So I think it's a good comparison,
01:13:20but we can't.
01:13:21It's very difficult to get two
01:13:22things that are exactly the same,
01:13:24but different, and so in that you know
01:13:26in fact that's impossible, right?
01:13:29So one thing I can say is that,
01:13:32at least for the signal in the
01:13:33way that David's looking at it,
01:13:35there isn't a change through the session
01:13:38in the circumstance when animals.
01:13:41Thirsty so there.
01:13:42Waters stated,
01:13:43and they're just choosing between
01:13:45the two rewards that might show that
01:13:48there's a big impact of the way that
01:13:50that these are is post ingestive
01:13:52effects of sucrose versus maltodextrin.
01:13:54So so at least on the face of it,
01:13:57based on the analysis of this signal
01:13:59there during the reward period,
01:14:01there's nothing obvious.
01:14:02Maybe there are other ways that
01:14:04he could look at the signals more
01:14:06carefully to see if there is feedback,
01:14:09because.
01:14:09And as you know,
01:14:11you know there's a big literature
01:14:13on how post ingestive impacts of
01:14:17food impact functioning within
01:14:19the striedl circuits that that
01:14:20we all know and love so much.
01:14:22So I I would,
01:14:23I wouldn't say there's not an interaction.
01:14:26I.
01:14:26I think there probably is,
01:14:27but at least for the signal
01:14:29that he's looking at,
01:14:29he didn't note anything obvious.
01:14:32Text.
01:14:35I have a quick question
01:14:36or maybe not so quick.
01:14:38The VP is a surprisingly large
01:14:42and heterogeneous structure and
01:14:44so I have a couple of questions.
01:14:47One is from your recordings.
01:14:49Can you determine the cell types
01:14:53that are responding based on their
01:14:56firing patterns or some kind of
01:14:58algorithms that let you know about,
01:15:00let's say cholinergic neurons
01:15:03versus other types of neurons?
01:15:06And then here on top and then
01:15:09the other question is around
01:15:11a sub sections of the VP.
01:15:13Whether you see these kinds of responses
01:15:15uniformly throughout the structure,
01:15:16or whether the anterior versus
01:15:18the posterior is more responsive
01:15:20to these value measurements.
01:15:24Those are great questions because it's
01:15:26known that as as many of you may know,
01:15:28the VP contains a lot of
01:15:30kinds of different neurons.
01:15:31It's it's mostly Gabaergic neurons,
01:15:34but there are neurons that release
01:15:36glutamate cholinergic neurons.
01:15:37It's a mix, and it's an area that
01:15:40doesn't have easily discernible
01:15:42boundaries necessarily.
01:15:43So when people work with it
01:15:45in in the rat and the road,
01:15:47it's a little bit difficult.
01:15:48So one way that David went about that that
01:15:51gets to your second point is he tried to.
01:15:54Positional has electrodes
01:15:55in sort of a central,
01:15:57not very anterior,
01:15:58not very posterior and operating medial,
01:16:00not very lateral area.
01:16:01So he could just be in the
01:16:03middle of the canonical BP,
01:16:04at least as described on Atlas is.
01:16:06So that's not necessarily
01:16:07what you might want,
01:16:09but he did not see variation based on
01:16:12electrode placement with his signals.
01:16:14Had he gone very far anterior
01:16:17posterior to search for that,
01:16:19he might have seen that because there
01:16:21is work emerging from other labs,
01:16:23including the Berridge lab.
01:16:25Showing that there the VP plays
01:16:27different roles in behavior when you
01:16:30are more anterior versus more posterior.
01:16:32So that's an important issue
01:16:34that that we've not looked at as
01:16:36far as cell types in the rat.
01:16:38So not having the the beauty of
01:16:40the transgenic mouse to let let us
01:16:43access different cell types easily,
01:16:45like Gabaergic cells,
01:16:46glutamatergic cells, etc.
01:16:47We can't say for sure.
01:16:50The area that he implanted his
01:16:52electrodes and his mostly Gabaergic.
01:16:55When we you there is a small
01:16:58a glutamatergic population,
01:16:59though,
01:16:59for example that people studied
01:17:01that has a critical behavioral role.
01:17:03When we use waveform shape and other
01:17:06neuro physiological characteristics to
01:17:07try to cluster neurons into different types,
01:17:10we get mostly,
01:17:11you know,
01:17:11big amorphous cluster that we presume
01:17:14is largely these canonical Gabaergic
01:17:16neurons and a smaller cluster that we
01:17:19guess could be the glutamatergic neurons,
01:17:21because it tends to signal very
01:17:24differently from the Gabaergic.
01:17:26Iran,
01:17:26so it doesn't show any of the same
01:17:29kinds of signals that I've just
01:17:30talked to you about, but we can't.
01:17:32And unless we do something
01:17:33you know more rigorous,
01:17:35something genetic like using
01:17:36viruses somehow in the rat to
01:17:38access the different populations
01:17:39or redo the work in the house.
01:17:41We really can't say for sure,
01:17:42and we've not even begun to think
01:17:45about or touch on the cholinergic.
01:17:48Aspects of this?
01:17:48I'm ashamed to admit in front of Marina,
01:17:52oh, that's OK. Are there any other questions?
01:17:56I don't see anything else in the track chat.
01:18:01If not one more please or I'm Joe.
01:18:07So I actually have to.
01:18:10I guess it's more related
01:18:12to technical questions.
01:18:13One was that when you used
01:18:17optogenetics that you injected.
01:18:19Inhibitory virus bilaterally.
01:18:22But you injected unilateral for the stimulus.
01:18:26So I was wondering why there is
01:18:28difference used by letter for
01:18:30inhibitory and Union letter for
01:18:33stimulus and the second one was.
01:18:35So all the dopamine signals were measured
01:18:38during the first trial for the
01:18:41second half of the experiments.
01:18:43And do you think there will be
01:18:45difference if you measure the
01:18:47signals during the free trials?
01:18:49Because Brett doesn't.
01:18:50No, what they're going to
01:18:52get when they press the.
01:18:54Or ignore spoke or pistol ever?
01:18:57Yes, those are both good questions
01:18:59I I think I would do that.
01:19:00The second one first and that is
01:19:02so in the in the forced trials.
01:19:04The rat doesn't know what reward it will get,
01:19:07so that's when we see these big signals.
01:19:10That are very different,
01:19:11but in the free trials,
01:19:12when the animal presses
01:19:14the lever for sucrose.
01:19:15He or she knows that that that it's
01:19:17about to drink sucrose in the in the port.
01:19:20So expectation already reduces
01:19:22that signal because he knows
01:19:24already what's going to happen,
01:19:26so that that's a great question,
01:19:28and one reason why we didn't look
01:19:30at those signals in great detail
01:19:32through the session is because
01:19:33by the time you get to the end of
01:19:35the session is really only enough
01:19:37data from the sucrose trials.
01:19:39The beginning of the session.
01:19:40There's really only enough
01:19:41data for the water trials,
01:19:43but it would be interesting
01:19:44still to show those.
01:19:45Think about those more and I
01:19:47think that's a nice question,
01:19:49and it might be interesting to look
01:19:50at the neural activity around the
01:19:52lever press for example so that there
01:19:54are other other time periods that
01:19:56I think I would be interested in.
01:19:58Knowing what are those neurons doing
01:20:00during those other time periods?
01:20:02The first question you asked is
01:20:04about the optogenetics procedure
01:20:05and so this is just something
01:20:07away that we've typically done it,
01:20:08and it's probably mostly out of convenience.
01:20:10So when you're activating neurons
01:20:12you usually can change the animal's
01:20:15behavior with.
01:20:15Optogenetics because it's such
01:20:17a strong and artificial approach
01:20:19usually can change the behavior by
01:20:22impacting the brain just on one side,
01:20:24and it's easier for the experimenter
01:20:26because then there's only one connection
01:20:28that they need to make with the.
01:20:29With the Rotary joint and all of this,
01:20:31but when you're inhibiting in a
01:20:33particular brain region a lot of
01:20:35times behavior is still not greatly
01:20:37impacted or even almost normal if the
01:20:40other side of the brain is still functioning,
01:20:42so to avoid a possible outcome like that.
01:20:46For inhibition,
01:20:47people often will do it bilaterally and try
01:20:50to inhibit in both both sides of the brain.
01:20:53So it's it's.
01:20:54It's really just a.
01:20:55Experimental practicality
01:20:58thank you so much, sure.
01:21:03OK. Well, I I want to thank
01:21:07you again for a great lecture.
01:21:09Thank you all for being here
01:21:11and for your great questions
01:21:12and let's thank Dr Janik again.
01:21:15And for those of you who are
01:21:16having who are trainees who
01:21:18are having lunch with her,
01:21:19you should have the the link to
01:21:22the other zoom room and I hope
01:21:26you'll move over to that room to
01:21:29talk with Doctor Genich further.
01:21:31Thanks I wanna thank.
01:21:32I want to thank you all so much.
01:21:33This was really a pleasure and
01:21:35thanks for your great questions.