WEBVTT 00:00:03.637 --> 00:00:06.517 This is the Convergent Science Network podcast. 00:00:08.557 --> 00:00:12.997 Leading researchers in the domain of neuroscience, brain theory and technology 00:00:12.997 --> 00:00:17.637 are interviewed by Paul Verschure and Tony Prescott. Okay. 00:00:18.457 --> 00:00:21.877 This is Paul Verschure with the Convergent Science Network podcast. 00:00:22.817 --> 00:00:28.637 And I'm here with Stefano Ferreira, who was one of our speakers in our summer school today. 00:00:28.637 --> 00:00:35.757 And Stefano you talked about logical neurons for logical reasoning what what does it exactly mean, 00:00:36.817 --> 00:00:39.517 yeah to me it's a 00:00:39.517 --> 00:00:47.617 first evidence I mean and at least in the neurophysiological literature on non-human 00:00:47.617 --> 00:00:57.397 primates that neurons in the brain are able to code for a logical property that is the 00:00:57.457 --> 00:01:01.817 one that is related to the transitive inference task. 00:01:02.637 --> 00:01:07.217 Okay, so we're talking monkeys here, primates, right? 00:01:07.657 --> 00:01:13.217 And so what you're trying to assess is, okay, what's the logical capability 00:01:13.217 --> 00:01:18.477 of the monkey as an organism and how much that can we recover in a neural response? 00:01:18.937 --> 00:01:21.597 So what's exactly the task that you were studying? 00:01:22.877 --> 00:01:28.257 Yeah, the task is named as a transitive inference task has been used in different 00:01:28.257 --> 00:01:31.857 forms in many types of animals. 00:01:32.537 --> 00:01:41.817 In a few words, it's the ability to conclude that A is higher than C after learning 00:01:41.817 --> 00:01:45.097 that A is higher than B and B higher than C. 00:01:46.037 --> 00:01:53.077 So to conclude that the first one is necessary to have a kind of relationship 00:01:53.077 --> 00:01:59.477 between all the values presented And then to be able to argument, 00:01:59.557 --> 00:02:03.777 to obtain more information than those provided at the beginning. 00:02:04.417 --> 00:02:08.617 Right, so it's like you have a sequence of magnitudes, let's say, 00:02:08.717 --> 00:02:12.557 and you learn to make inferences about these relationships. 00:02:12.717 --> 00:02:19.257 Correct. And you also drew a parallel to the social life of certain kinds of primates. 00:02:19.977 --> 00:02:23.817 Was it just to introduce the topic, or do you really see this as reflecting 00:02:23.817 --> 00:02:28.317 a need that they have in the wild? 00:02:28.637 --> 00:02:32.817 No, it was a way to introduce, but it's also true that in the other animals 00:02:32.817 --> 00:02:38.877 where the property of transit inference has been shown, this is always stronger, 00:02:39.197 --> 00:02:44.937 I mean, more evident in animals that are used to live in such organized groups. 00:02:45.877 --> 00:02:53.517 Okay, so there might be a correlation. There's one study that compared two types of birds. 00:02:54.565 --> 00:03:02.985 Very similar in the species, but since one of the two lives normally in the 00:03:02.985 --> 00:03:08.245 large group while the second one is as a group very reduced, 00:03:08.625 --> 00:03:15.365 the evidence in this study was that the transitive inference properties was 00:03:15.365 --> 00:03:22.505 more evident in the all-in-one living in large organized groups. Okay. 00:03:23.005 --> 00:03:29.565 Okay. So now the specific task that the monkey was exposed to was essentially 00:03:29.565 --> 00:03:33.425 using different kinds of shapes, fairly abstract shapes. 00:03:34.065 --> 00:03:39.625 And you would present pairs of shapes. And then one of them would lead to reward 00:03:39.625 --> 00:03:43.245 in the training phase, right? It would give reward. The second one would not give reward. 00:03:43.985 --> 00:03:49.025 And then in the next pair, you would take, let's say, the second pattern that 00:03:49.025 --> 00:03:51.185 didn't get reward, you would present it with a third pattern. 00:03:51.185 --> 00:03:55.365 And then the second pattern would get reward and the third one wouldn't get 00:03:55.365 --> 00:03:59.245 reward etc and that's how you go down the chain so we have a certain specific sequence, 00:04:00.105 --> 00:04:06.365 and then in the test phase you present the animals now with randomly selected 00:04:06.365 --> 00:04:11.025 pairs of the whole sequence that they were trained on including those trained 00:04:11.025 --> 00:04:15.765 exactly and then they have to choose the one that would give them more reward than the other, 00:04:16.785 --> 00:04:19.345 that's the whole task so. 00:04:20.985 --> 00:04:25.285 So what do you really observe when monkeys perform this task? 00:04:25.545 --> 00:04:27.065 How good are monkeys at this task? 00:04:28.025 --> 00:04:32.325 Actually, surprisingly, the monkey were very good on performing, 00:04:32.465 --> 00:04:37.345 actually on learning the general idea of the task and acquiring the strategy. 00:04:38.665 --> 00:04:45.505 After less than three weeks of initial training, meaning both monkeys we used 00:04:45.505 --> 00:04:49.085 were able to play with figures. 00:04:50.497 --> 00:04:58.897 To recognize by trial and error that there is an imposed series that the experiment, 00:04:59.857 --> 00:05:07.857 presented to them, and then to use this information also at the moment of the test when, 00:05:08.897 --> 00:05:13.277 novel figures, never matched figures during training, are presented, 00:05:13.297 --> 00:05:18.137 like concluding in a sequence from A to F who is the, 00:05:19.377 --> 00:05:21.877 rewarded one between B and E. 00:05:23.157 --> 00:05:29.677 But now, in the training, what kind of, if we talk about a difference in reward magnitude, 00:05:30.077 --> 00:05:35.237 the amount of juice you get, if you go to the stimulus, the pattern with the 00:05:35.237 --> 00:05:39.617 highest reward and lowest reward, what's the difference in reward that they 00:05:39.617 --> 00:05:40.337 got, what kind of magnitude? 00:05:40.337 --> 00:05:46.457 We never introduced this variable to the task that is actually interesting and 00:05:46.457 --> 00:05:50.997 could be explored in the future, at least this is our hope. 00:05:52.097 --> 00:05:58.717 But to say differently, the amount of reward was always the same. 00:05:58.817 --> 00:06:04.937 For each pair, there was a rewarded symbol and a rewarded one. 00:06:04.937 --> 00:06:14.497 So there was never a symbol with a strong amount of reward and the other one with less. 00:06:14.757 --> 00:06:20.117 So the comparison was always relative within each pair. 00:06:20.337 --> 00:06:26.517 And the series was created because the different pairs were presented consequently. 00:06:27.137 --> 00:06:30.777 Right, exactly. Exactly. So that would mean in your test phase you cannot combine 00:06:30.777 --> 00:06:36.057 all possible patterns because then you would have two patterns that in that 00:06:36.057 --> 00:06:39.097 sequence would never have been rewarded or would always have been rewarded. 00:06:39.817 --> 00:06:45.897 No, in the test all the comparisons are compared. I mean all the possible pairs are compared. 00:06:46.357 --> 00:06:51.417 Okay. So that would mean you might present patterns that in the combination 00:06:51.417 --> 00:06:53.437 would never have been rewarded in any way. 00:06:54.791 --> 00:06:58.571 Well, that's not true. 00:06:58.631 --> 00:07:06.971 No, it's not true because the only one figure on a sequence of six using letter is E. 00:07:09.291 --> 00:07:16.131 F is F. The F is never rewarded during the learning. That's right. Exactly. 00:07:16.351 --> 00:07:25.071 While A is always rewarded. So, one easy conclusion is that it's very simple 00:07:25.071 --> 00:07:34.371 for the monkey to conclude when A and E are present in the test, 00:07:34.491 --> 00:07:40.311 but our data show that it's not true, it's not just a simple relationship between 00:07:40.311 --> 00:07:45.211 a real world history and logical assumptions. 00:07:46.911 --> 00:07:51.151 And what's the question again? No, my question is just about the test, 00:07:51.291 --> 00:07:52.871 whether there's not a bias in the test. 00:07:52.991 --> 00:07:56.631 Because like you say now, stimulus F is never rewarded. 00:07:56.931 --> 00:08:01.731 So in some sense, there are probe trials where you have pairs that include stimulus 00:08:01.731 --> 00:08:05.991 F, which would be much easier to evaluate 00:08:05.991 --> 00:08:10.331 than if you would have any other combination of stimuli from A to E. 00:08:10.331 --> 00:08:21.631 Yeah, but still having E and F that are adjacent in the pair presentation during learning is sometimes, 00:08:21.911 --> 00:08:28.351 because the performances are probabilistic, sometimes more difficult than comparing B and F. 00:08:28.531 --> 00:08:32.311 Even if F is always there and F was never there. 00:08:33.291 --> 00:08:38.651 That's a surprising effect that you found, right? If you now do these test trials 00:08:38.651 --> 00:08:44.951 and we present combinations of stimuli, if you would have A and B... 00:08:46.329 --> 00:08:51.489 In terms of reaction time, the monkey might take longer to make a decision than 00:08:51.489 --> 00:08:53.249 if you present A or A and E. 00:08:53.769 --> 00:08:56.909 So how do you explain that? This is right. 00:08:57.749 --> 00:09:09.029 One was suspecting that those figures more presented during learning are the ones more easily coded. 00:09:09.669 --> 00:09:12.669 And this is not what emerged from the behavioral data. 00:09:12.669 --> 00:09:20.309 The behavioral data shows that comparing figures that are more far located in 00:09:20.309 --> 00:09:24.909 the series is absolutely easier for the animal. 00:09:26.509 --> 00:09:32.669 This is quite a surprise. According to some people that presented the model 00:09:32.669 --> 00:09:39.509 in the literature, this is probably due to the noisy representation of each symbol in the series. 00:09:39.509 --> 00:09:45.069 And then comparing two very close symbols that are noisy is much more complicated 00:09:45.069 --> 00:09:50.349 because of the noise when that comparing two symbols that are far located. 00:09:50.629 --> 00:09:54.069 But could it not be an interference effect of multiple memory systems? 00:09:54.209 --> 00:09:55.049 It's a certain way. It's an interference. 00:09:55.509 --> 00:10:02.069 I think interference and noise is the same because noisy means that the symbol 00:10:02.069 --> 00:10:09.029 is not very well coded when you compare the code to the very close to them. 00:10:09.509 --> 00:10:13.829 To the other symbol very close to them in the series. So if there is no noise, 00:10:13.869 --> 00:10:15.949 it means that they are very well classified. 00:10:17.089 --> 00:10:20.449 So if they are not very well classified, there is interference. 00:10:21.349 --> 00:10:25.329 Sure, but don't you agree that noise is a bit of a dissatisfactory explanation 00:10:25.329 --> 00:10:27.049 of variability in nature? 00:10:27.229 --> 00:10:30.129 Because basically we're not actually explaining anything. We just say, 00:10:30.189 --> 00:10:32.729 well, there's some uncontrollable variability here. 00:10:32.849 --> 00:10:40.489 Well, you could also argue that maybe the strange thing is If I'm being trained on a certain pair AB, 00:10:40.769 --> 00:10:47.269 right, and I'm being tested on ANC or ANB, and I compare my reaction time or 00:10:47.269 --> 00:10:49.029 my accuracy, it's worse than ANB. 00:10:49.329 --> 00:10:54.049 So I could argue, well, maybe since I'm instantaneously confronted with ANB 00:10:54.049 --> 00:10:58.429 in the training set, it's also a pattern association mechanism that comes into 00:10:58.429 --> 00:11:00.609 play. There's a perceptual mechanism that comes into play. 00:11:00.669 --> 00:11:03.769 It says, well, maybe ANB is, let's see, one category. 00:11:04.429 --> 00:11:07.269 Okay. And now I get two competing players. 00:11:08.241 --> 00:11:12.021 Memory-dependent processes, one telling you something about, 00:11:12.101 --> 00:11:17.101 let's say, the magnitude order, the order of presentation, and the interference 00:11:17.101 --> 00:11:20.281 comes from a process that not all these things belong together, let's say. 00:11:20.961 --> 00:11:24.221 So that you really have interference between specific cognitive processes as 00:11:24.221 --> 00:11:28.401 opposed to a nonspecific noise with a decision-making stage. 00:11:28.701 --> 00:11:32.701 I agree. I actually believe that there are different levels of representation. 00:11:33.061 --> 00:11:38.201 One is that the single symbols are represented in the brain. 00:11:38.341 --> 00:11:41.441 The other one is that the pair starts to be represented. 00:11:42.041 --> 00:11:47.341 Since the figure B is present in both A, B, and B, C, 00:11:47.441 --> 00:11:54.121 and in the two cases the probability of reward is 50%, then the representation 00:11:54.121 --> 00:11:59.341 in the pair need to be, because it's much more convenient, more noisy than the 00:11:59.341 --> 00:12:01.801 representation limitation on the single item. 00:12:01.901 --> 00:12:10.421 So at the time when the monkey is required to answer which one is linked to the reward. 00:12:11.141 --> 00:12:15.141 Of the never presented, the never experienced pair, for the monkey it's very 00:12:15.141 --> 00:12:19.141 easy because there was no code, no previous code assigned to them. 00:12:19.301 --> 00:12:27.461 So there's no ambiguity because it's not part of the A, B, B, C conflict, it's a B, D. 00:12:27.461 --> 00:12:36.981 So, for the monkey the job is easier, but this is my point to try to figure out what's, 00:12:37.521 --> 00:12:46.241 the category of the pair based on the, even in the conflicting experience, 00:12:46.541 --> 00:12:48.141 but based on previous experience. 00:12:48.501 --> 00:12:51.321 Because one interpretation could also be you might 00:12:51.321 --> 00:12:54.421 interpret a single stimulus presentation is let's 00:12:54.421 --> 00:12:57.421 say an episode a memory episode right in which everything comes 00:12:57.421 --> 00:13:00.441 together it's okay i saw these two patterns together and now 00:13:00.441 --> 00:13:04.221 i see some other patterns together so then those episodes 00:13:04.221 --> 00:13:09.841 might be then chained together in the sequence right so and that comparisons 00:13:09.841 --> 00:13:14.481 between episodes and within episodes are actually very different processes that 00:13:14.481 --> 00:13:19.321 that's essentially what you're what you're seeing yeah okay but then you made 00:13:19.321 --> 00:13:22.001 you made also That's an important point in terms of, 00:13:22.061 --> 00:13:27.541 because for you, this data is pointing to an ability for logical operations. 00:13:28.401 --> 00:13:31.621 Now, for that, you must exclude interpretation that would say, 00:13:31.681 --> 00:13:32.581 well, this is just reward. 00:13:33.661 --> 00:13:37.461 Dependent on the reward that you receive, you form certain associations between 00:13:37.461 --> 00:13:41.081 events, and it's just a very direct recall of these associations. 00:13:41.321 --> 00:13:44.301 So how can you exclude this interpretation? 00:13:46.081 --> 00:13:54.641 The approach that we use that is referred to other similar approaches in the 00:13:54.641 --> 00:14:01.881 literature is that we ask the monkey to learn two different chains. 00:14:03.308 --> 00:14:09.808 And in this case, the history or reward is distributed differently in the two chains. 00:14:11.268 --> 00:14:15.488 And then at the end of the learning of the two separate chains, 00:14:15.568 --> 00:14:23.028 we asked the monkey to link the two chains by providing them the relative value of the extremes. 00:14:23.988 --> 00:14:29.848 That means C in the ABC chain and D in the DEF chain. 00:14:30.628 --> 00:14:41.188 And the evidence shows that the monkey is still expressing behaviorally the same, 00:14:42.348 --> 00:14:46.088 performance than when the chain is unique. 00:14:46.328 --> 00:14:55.868 Even if the reward is not any more uniformly distributed to explain the relative 00:14:55.868 --> 00:15:00.028 value only based on these attributes. 00:15:00.408 --> 00:15:07.088 So we conclude that probably the monkey is not only basing their conclusion 00:15:07.088 --> 00:15:13.448 on the reward, but it's also using the information provided previously differently. 00:15:13.788 --> 00:15:19.108 And in this kind of chaining test, so let's say we have two sequences, 00:15:19.168 --> 00:15:25.728 like you say, ABC and with DEF, And now suddenly I present the monkey with B 00:15:25.728 --> 00:15:28.188 and E as a pair, right, to evaluate. 00:15:29.908 --> 00:15:35.048 Does the monkey immediately respond to that correctly or they have to learn 00:15:35.048 --> 00:15:38.128 that this is now a relevant question? 00:15:38.448 --> 00:15:42.708 Unfortunately, I have no answer to this question because we haven't looked to 00:15:42.708 --> 00:15:45.008 this data, this level of detail. 00:15:45.008 --> 00:15:50.888 Detail, but I know from all of our data that during the task, 00:15:50.928 --> 00:15:54.768 the monkey continue to learn, but this is normal. Of course. 00:15:55.228 --> 00:16:02.248 So I don't know if in the chain experiment, there is already the evidence of 00:16:02.248 --> 00:16:05.208 the link between the two. 00:16:05.508 --> 00:16:08.448 Right, exactly. So your point is, look, that they can do this, 00:16:08.508 --> 00:16:12.068 that they perform well in the chain experiment experiment would 00:16:12.068 --> 00:16:15.288 require some form of reasoning let's say deliberation which 00:16:15.288 --> 00:16:18.348 said well no i had to have this sequence abc and ah 00:16:18.348 --> 00:16:25.248 this was the other sequence df now e was later than b in in my in the learning 00:16:25.248 --> 00:16:31.588 in the learning phase so now i know i have to choose for b right but i you could 00:16:31.588 --> 00:16:35.308 also argue that maybe again this is a perceptual learning process because now 00:16:35.308 --> 00:16:38.048 uh the first time i I see B and E, 00:16:38.128 --> 00:16:41.868 so the two exemplars from the two different sequences, 00:16:42.068 --> 00:16:45.908 I don't know what to do, let's say, but they're presented together, 00:16:46.168 --> 00:16:51.528 they're now conjunctive, and dependent on my choice, I get reward or no reward. 00:16:52.128 --> 00:16:53.808 So I could learn a new contingency. 00:16:54.688 --> 00:16:57.268 So how could you exclude that interpretation? Yeah. 00:16:58.231 --> 00:17:05.491 Again, I'm not able to exclude without looking at the first trials of the test. 00:17:06.371 --> 00:17:13.391 But I bet also that you do agree that this is no simple sensory motor transformation 00:17:13.391 --> 00:17:15.771 of the symbol in the response. 00:17:16.031 --> 00:17:23.211 There is something that is more than associating symbols to response. 00:17:23.211 --> 00:17:29.231 So when the overall effect of the symbolic distance effect, 00:17:29.431 --> 00:17:34.551 that is the performance that changes with the distance between the symbols that 00:17:34.551 --> 00:17:38.371 are supposed to be located in the mental space differently, 00:17:38.811 --> 00:17:44.911 and this behavioral effect is still there after linking two different chains, 00:17:45.111 --> 00:17:48.331 I think it's a good argument in favor of. It's not a conclusion. 00:17:49.531 --> 00:17:54.411 No, that's indeed interesting, right? So we have these alternative interpretations, 00:17:54.671 --> 00:17:56.291 and we can come back to that later. 00:17:56.471 --> 00:18:03.491 But indeed, what you subsequently showed is that the two main phenomena you 00:18:03.491 --> 00:18:06.931 then looked at was what you called symbolic distance and serial distance, 00:18:07.031 --> 00:18:08.011 and you compared the two. 00:18:08.191 --> 00:18:11.911 So what's the difference? What's a symbolic distance and what is serial distance? 00:18:12.051 --> 00:18:16.051 How are they different? Yeah, the symbolic distance is defined as the effect 00:18:16.051 --> 00:18:22.251 in the behavioral measure, that is the measures that are the reaction time and 00:18:22.251 --> 00:18:26.291 the performance, that is essentially better. 00:18:26.531 --> 00:18:31.751 I mean, the performance increase and the reaction time is reduced when the symbols 00:18:31.751 --> 00:18:34.611 to be compared are more distant each other. 00:18:34.611 --> 00:18:43.011 So, comparing B and E is easier than comparing the learned one B and C pair. 00:18:43.431 --> 00:18:46.171 So this is the symbolic distance effect. 00:18:47.491 --> 00:18:51.271 It's a very strong effect that has been previously described in humans too, 00:18:51.371 --> 00:18:58.631 and it's common to other quantity comparison like in the numerical literature. 00:18:58.631 --> 00:19:05.211 So again, in the numerical literature, when a subject is asked to compare between 00:19:05.211 --> 00:19:13.991 a number that is five as a reference in the first 10 and six or five and eight, 00:19:14.151 --> 00:19:21.491 normal subjects have more difficulty on concluding the six is higher than five 00:19:21.491 --> 00:19:23.071 than eight is higher than five. 00:19:25.004 --> 00:19:35.044 The second effect is the serial position effect. That is the evidence that in the learned series, 00:19:35.384 --> 00:19:47.084 the different symbols are not represented equally, but they are some way organized around a center. 00:19:47.244 --> 00:19:52.424 This is in our case the letter C that stays in the middle of the series. 00:19:52.424 --> 00:19:56.044 But before we go to that point, 00:19:56.604 --> 00:20:07.044 if we look at symbolic distance, and we now take the case that we're linking 00:20:07.044 --> 00:20:09.464 two sequences together, right? Right. 00:20:10.344 --> 00:20:18.924 Now, the only sticking point I have is that if I have a three-element sequence, ABC, 00:20:20.184 --> 00:20:25.124 that the combination C with anything from the second sequence is now an inconsistent 00:20:25.124 --> 00:20:27.064 pairing compared to everything else. 00:20:27.424 --> 00:20:29.724 Because C was never rewarded. Correct. 00:20:30.564 --> 00:20:38.104 So how is that exception handled in terms of the performance of the monkey? 00:20:40.742 --> 00:20:43.702 Because it's like an exception that they must handle, right? Yeah. 00:20:45.342 --> 00:20:52.062 It's true that you are referring to local phenomena, while unfortunately we 00:20:52.062 --> 00:20:56.242 analyze our behavioral data at the global level. Okay, okay, all clear. 00:20:56.642 --> 00:21:06.302 But I think it's important to remember that what is one of the strategies that 00:21:06.302 --> 00:21:08.622 I suppose our animals were able to learn, 00:21:08.622 --> 00:21:13.362 that there is no absolute value of every symbol presented. 00:21:13.782 --> 00:21:21.202 It's the same at the time when B is presented for the first time with A, 00:21:21.442 --> 00:21:28.682 B is never reinforced in the block design of the learning, but then right after, 00:21:28.802 --> 00:21:31.942 B is presented together to C, and now it's B reinforced. 00:21:31.942 --> 00:21:38.902 So, for the message that is provided to the animal is that there's a warning 00:21:38.902 --> 00:21:47.002 that the value of the symbol in terms of probability of getting a reward could be changed shortly. 00:21:48.362 --> 00:21:52.602 So this is probably the same that happens with C because the monkey already 00:21:52.602 --> 00:21:56.662 know that this is possible and the two chain experiment, 00:21:56.902 --> 00:22:02.342 even if the consolidation and the training was very long, longer than the A-B 00:22:02.342 --> 00:22:06.082 training of a single series. 00:22:06.642 --> 00:22:13.342 The monkey knows that the C is possible for the C to be categorized differently. 00:22:13.702 --> 00:22:17.782 And this is what happens with the linking pair. At a certain point, 00:22:18.002 --> 00:22:25.742 I'm telling to the monkeys, C is also a winner only when presented with D. Right, exactly. 00:22:26.122 --> 00:22:31.822 Okay. Okay, so then you're saying in the test phase itself, this might then 00:22:31.822 --> 00:22:35.742 mitigate this idea that it's an exception, that the monkey is just processing 00:22:35.742 --> 00:22:40.262 it as any other sequence because it's, well, look, I might not get reward now, but maybe in the future. 00:22:40.562 --> 00:22:45.982 This is what I believe, yeah. Okay, all right. So then, for symbolic distance, 00:22:46.402 --> 00:22:52.702 it's a term used also in the psychological literature, but in your task, 00:22:52.882 --> 00:22:57.002 is it reasonable to really speak of a symbol as you're using these patterns? 00:22:57.002 --> 00:23:02.082 And the monkey must also really react to the patterns as they're localized in space, right? 00:23:02.122 --> 00:23:05.482 They must really touch the screen at a certain X, Y position. 00:23:05.722 --> 00:23:10.042 And you could argue, well, a symbol might be a difficult concept to define. 00:23:10.202 --> 00:23:14.242 But the one thing is that it is an internal representation that is independent 00:23:14.242 --> 00:23:16.022 of the actual sensory state. 00:23:19.842 --> 00:23:23.822 I think that the symbols are very large. 00:23:24.722 --> 00:23:31.702 They don't need to be precisely recognized, I mean, and not even the position 00:23:31.702 --> 00:23:34.342 on the touch screen need to be precisely coded. 00:23:34.982 --> 00:23:38.822 For the monkey, the response is left and right, it's the same of a key press. 00:23:40.713 --> 00:23:48.913 So, what I mean more is that, is it fair to call the pattern that you use? You project a pattern. 00:23:49.313 --> 00:23:53.513 Yeah, but the monkey, if you remember the video, the monkey is looking for the figure. 00:23:53.653 --> 00:23:58.273 It's not stay still on the pattern. 00:23:58.773 --> 00:24:03.373 I mean, it doesn't look at the pair. It looks for this figure. 00:24:03.733 --> 00:24:10.473 Right. And then there is another behavioral data that we haven't discussed. It is the eye movement. 00:24:10.713 --> 00:24:13.493 We know that 00:24:13.493 --> 00:24:16.453 if the first figure that is 00:24:16.453 --> 00:24:19.333 4V8 is the wrong one then he moves the 00:24:19.333 --> 00:24:22.333 eye to the next one and then decide 00:24:22.333 --> 00:24:25.353 to move the arm so there are other aspects of 00:24:25.353 --> 00:24:29.333 the behavior that need to be further explored okay what 00:24:29.333 --> 00:24:35.313 I'm after here maybe it's a completely irrelevant semantic concern I'm expressing 00:24:35.313 --> 00:24:41.093 but you want to measure something called symbolic distance but what you're using 00:24:41.093 --> 00:24:46.613 is also a spatially organized task with visual patterns that in themselves cannot be considered symbols. 00:24:47.953 --> 00:24:50.273 So what makes it symbolic? 00:24:52.673 --> 00:24:59.233 Okay, I could agree that the term symbolic is not proper but for me it's the 00:24:59.233 --> 00:25:03.393 way to refer to something that is not in the literature. 00:25:03.873 --> 00:25:09.233 Okay. And I agree with you that probably a symbol is never created in this kind 00:25:09.233 --> 00:25:11.613 of strategy that we are imposing to the animal. 00:25:12.473 --> 00:25:19.953 Because just to be complete, often in the animal study and even in human study, 00:25:20.853 --> 00:25:26.213 the symbol used before testing are always the same in a way that the subject 00:25:26.213 --> 00:25:29.493 creates a symbolic representation of the symbol. Exactly. 00:25:29.713 --> 00:25:34.493 And actually for some of the people that discuss a lot on the behavioral relevance 00:25:34.493 --> 00:25:39.133 of transitive inference, this is a necessary step. 00:25:39.373 --> 00:25:47.293 Right, okay. But in our case, we want to avoid that the monkey was just using a well-coded algorithm, 00:25:48.240 --> 00:25:54.100 symbol to conclude on the comparison between two symbols. 00:25:57.000 --> 00:26:03.440 I think this is just maybe not the right one solution, but it's a way to reduce 00:26:03.440 --> 00:26:05.620 our... No, no, look, I completely get it, sure. 00:26:06.400 --> 00:26:13.260 The number of factors that are there, because remember that then we try to correlate 00:26:13.260 --> 00:26:17.380 this behavioral data with a neural code. 00:26:17.380 --> 00:26:22.200 And if we were very too many factors at the same time. 00:26:22.420 --> 00:26:30.180 Of course. No moreover it might so as an operational definition of a symbol 00:26:30.180 --> 00:26:34.380 I think it's completely defendable but we just have to be clear about what we're looking at. 00:26:34.380 --> 00:26:38.780 So I agree with you that the symbolic distance in this case is not properly. 00:26:39.520 --> 00:26:46.500 But now what you found is that if you look at the performance of the animals, 00:26:46.600 --> 00:26:49.500 so you look at the errors they make and their reaction times, 00:26:49.740 --> 00:26:55.260 then you see that they perform better on patterns or symbols, 00:26:55.400 --> 00:27:00.020 as we now can call them, that are in the middle of the sequence as opposed to 00:27:00.020 --> 00:27:03.220 those that are at the beginning or the end of a sequence, right? 00:27:05.340 --> 00:27:15.000 No, the performance is better for the anchor points, even for the one that includes 00:27:15.000 --> 00:27:17.920 the never enforce it. Okay. 00:27:18.060 --> 00:27:21.860 So the beginning and the end. The beginning and the end. So how big is that modulation? 00:27:22.780 --> 00:27:28.120 In terms of difference in percentage of performance, it's I think twice. 00:27:29.200 --> 00:27:33.940 So it's a big difference. It's a big difference in the two animals that we test. Sure. 00:27:35.231 --> 00:27:38.671 But how do you explain that or what does that mean? 00:27:38.911 --> 00:27:45.911 I think that this is a partial evidence that the different symbols are organized, 00:27:46.431 --> 00:27:53.791 relatively to each other around the two anchor points that are the A and F of the sequence, 00:27:54.431 --> 00:27:58.391 and maybe also the central one that's the letter C. 00:27:59.371 --> 00:28:05.011 And at least in the human literature, This is a partial evidence that a line, 00:28:05.091 --> 00:28:10.811 a mental line is used because sometimes people refer that to solve the task, 00:28:10.851 --> 00:28:16.671 they do explore mentally the line by moving from the extremes toward the center 00:28:16.671 --> 00:28:20.971 to find the item that is required to be compared. 00:28:21.611 --> 00:28:27.191 So if you interpret this as arranging now these symbols on the line, 00:28:27.331 --> 00:28:30.751 that would basically mean that okay i'm exposed 00:28:30.751 --> 00:28:33.611 to the task i know now what the beginning and 00:28:33.611 --> 00:28:36.891 the end of my line is and then now 00:28:36.891 --> 00:28:39.991 it's like i build a chain right now i can stick the elements in 00:28:39.991 --> 00:28:42.631 that on that line is that correct that would be the 00:28:42.631 --> 00:28:46.831 idea yes but that's what it was interesting of course that the two anchor points 00:28:46.831 --> 00:28:52.131 are in their quality radically different because that this is a surprise it's 00:28:52.131 --> 00:28:57.751 still we have a very similar behavior that means that the the behavior is It's 00:28:57.751 --> 00:29:00.971 not just explained by the reward. That's exactly right. 00:29:01.371 --> 00:29:04.091 So do you see this as a form of latent learning for the endpoint? 00:29:04.431 --> 00:29:06.951 So this is learning without an explicit instruction? 00:29:08.691 --> 00:29:12.411 Could be. Yeah. It could be a possibility. Yeah. 00:29:12.491 --> 00:29:17.231 Could it also be a violation? I named this a strategy because I think that the 00:29:17.231 --> 00:29:22.831 monkey needs to have a solution because every day he's introduced in the lab. 00:29:23.031 --> 00:29:26.291 He doesn't know which figure will be used. use it but the 00:29:26.291 --> 00:29:30.371 monkey knows that you have to use this new figure to 00:29:30.371 --> 00:29:33.931 figure out uh how they 00:29:33.931 --> 00:29:40.451 are located relatively each other right so the it could be that just the monkey 00:29:40.451 --> 00:29:49.731 normally I mean spontaneously adapt uh to this uh allocation in this in the 00:29:49.731 --> 00:29:53.171 mental space of of the symbols presented. Right, exactly. 00:29:54.111 --> 00:30:01.131 Without any control. But now, could you argue that this is like a segmentation 00:30:01.131 --> 00:30:03.531 of the task? Would you accept that? 00:30:06.073 --> 00:30:07.053 Or the task domain? 00:30:10.393 --> 00:30:15.853 Yeah, let's say differently. I have no argument to conclude that it's not segmented. 00:30:16.633 --> 00:30:20.433 So I think that it's in favor of a segmentation. Right. 00:30:20.933 --> 00:30:25.113 Because what is interesting then is that sometimes the monkey is saying, 00:30:25.193 --> 00:30:26.973 okay, I have a problem to solve. Yeah. 00:30:28.213 --> 00:30:33.633 These experimenters only show me always a subset of the task I must solve. 00:30:33.773 --> 00:30:35.073 I have to figure it out myself. 00:30:36.073 --> 00:30:38.533 So I have to learn the boundaries. I have to figure out what are the boundaries 00:30:38.533 --> 00:30:42.673 of my task, right? Where do the relationships start and where do they stop? 00:30:43.213 --> 00:30:46.693 And in the performance, you see that actually a monkey is able to do that and 00:30:46.693 --> 00:30:49.673 also to draw very strict boundaries around that. 00:30:51.253 --> 00:30:57.013 But the strange thing is that this almost looks like a form of perceptual learning 00:30:57.013 --> 00:31:03.653 in the sense that it is an acquisition of knowledge about the task without explicit 00:31:03.653 --> 00:31:07.293 instruction because the beginning gets reward and the end gets nothing, right? 00:31:08.033 --> 00:31:15.573 So do you see that as a separate kind of learning process that imposes boundaries 00:31:15.573 --> 00:31:20.233 on this local structuring of the task or do you see it as an integrated process? 00:31:21.913 --> 00:31:34.533 I believe that both are present because the assumption that A is always reinforced And F is never, 00:31:34.733 --> 00:31:40.813 it's not easy to explain why A and B is so difficult and A and F is so difficult. 00:31:41.153 --> 00:31:49.013 So I think that there are two interfering process that works together. 00:31:49.233 --> 00:31:55.053 One is the tentative of the monkey to generalize, to organize the item according 00:31:55.053 --> 00:31:58.133 to the general rule that has been introduced by the experiment. 00:31:58.133 --> 00:32:03.693 The second one is that there is a conflict on the representation that is created. Okay. 00:32:04.313 --> 00:32:09.973 And then, so now for this symbolic distance, which is the magnitude relationships 00:32:09.973 --> 00:32:16.473 that we are being trained on, like our binary relationship, then we have the serial order effect. 00:32:16.753 --> 00:32:20.233 Yeah. Right? How is the serial order effect or the serial distance effect different 00:32:20.233 --> 00:32:21.833 from the symbolic distance effect? 00:32:21.833 --> 00:32:30.573 The serial order is the phenomenon that we just described, that is the fact that the. 00:32:31.413 --> 00:32:39.893 Learned pair are not coded similarly with the U shape described, 00:32:40.273 --> 00:32:50.333 while the symbolic Symbolic distance is the relationship with the distance of the symbol compared, 00:32:50.673 --> 00:32:54.733 in particular for never-matched symbols during the learning, 00:32:54.933 --> 00:33:00.793 so that could be named novel pair. 00:33:03.598 --> 00:33:07.938 So, with that in mind, you started to look at the neural substrate. Yeah. 00:33:08.658 --> 00:33:15.138 And you record it in premotor cortex? It's prefrontal cortex this time. 00:33:15.658 --> 00:33:20.198 Prefrontal? I present today the data from prefrontal cortex. 00:33:20.278 --> 00:33:22.778 We do also have data from premotor. 00:33:22.858 --> 00:33:29.618 The reason is that in the literature, that is mostly in the human literature, 00:33:29.778 --> 00:33:36.918 using imaging data, the evidence is that some of the area of the cortex, 00:33:37.158 --> 00:33:43.578 essentially the prefrontal cortex, the premotor cortex and the posterior parietal 00:33:43.578 --> 00:33:49.078 cortex are involved to the transitive inference task. 00:33:49.298 --> 00:33:54.618 You know that the functional MRI doesn't provide the details to conclude what is the mechanism. 00:33:54.618 --> 00:34:00.238 So we only know that there are areas that are statistically activated while 00:34:00.238 --> 00:34:10.758 the subjects are tested in a similar task to the one that I use in the animal. 00:34:11.038 --> 00:34:15.378 So what did you observe recording in the frontal cortex? 00:34:15.378 --> 00:34:25.098 The observation that I think is very straightforward is that we have been able to find neurons. 00:34:25.958 --> 00:34:34.998 That are modulated in a way similar, another way is to say they are correlated 00:34:34.998 --> 00:34:40.858 to the behavioral effect that our animals are able to show. 00:34:40.858 --> 00:34:47.398 So there are neurons modulated for the symbolic distance, other neurons are 00:34:47.398 --> 00:34:49.198 modulated for the serial position, 00:34:49.558 --> 00:34:59.018 and a subpopulation of them, when analyzed at a single neuron, are modulated by both. 00:35:00.222 --> 00:35:11.022 I specified single neuron level because when the average activity of all the neurons task-related, 00:35:11.582 --> 00:35:19.222 are further investigated, then what emerges is that the population of neurons, 00:35:19.302 --> 00:35:24.482 which is about half of the total population we recorded with our movable electrode 00:35:24.482 --> 00:35:31.762 in the prefrontal cortex, is able to code both for the symbolic distance and 00:35:31.762 --> 00:35:32.902 the serial position effect. 00:35:33.182 --> 00:35:38.062 That is quite important because the two phenomena could be separated in the 00:35:38.062 --> 00:35:40.262 brain as used differently, 00:35:40.782 --> 00:35:46.922 by whatever code the brain is using to solve the task, 00:35:47.082 --> 00:35:54.502 while being able to demonstrate that the same population is able to use both 00:35:54.502 --> 00:36:00.302 signals means that probably the one signal is dependent on the other. Okay. 00:36:00.642 --> 00:36:07.502 But now, so the numbers you listed were about 50% responsive for serial position. 00:36:08.462 --> 00:36:13.502 40% for symbolic distance, and then there were a few neurons across these two 00:36:13.502 --> 00:36:15.262 pools that were responsive to both. 00:36:15.462 --> 00:36:19.722 Yeah. So like 30% or 20% was something? It's 20% of both. 00:36:20.002 --> 00:36:26.102 So, but now how do we know that this actually scales up? because what you really 00:36:26.102 --> 00:36:29.342 measured is you present a monkey now with a pair of patterns, 00:36:29.422 --> 00:36:32.682 let's say B and C or A and B. 00:36:34.142 --> 00:36:38.062 You record a neuron and now you basically sort all your trials. 00:36:38.122 --> 00:36:44.382 And so, okay, let's put all the trials together where I presented C as my first pattern to the left. 00:36:44.902 --> 00:36:47.562 And now I'm going to look at, okay, how do my neurons respond? 00:36:47.782 --> 00:36:52.362 And you will find the sensitivity of neurons to, let's say, C in the first position, 00:36:52.502 --> 00:36:55.642 right? And then you might find another neuron that says, okay, 00:36:55.702 --> 00:36:58.102 I like B in the second position, etc. 00:37:00.502 --> 00:37:06.062 But that basically at this point only tells us that we are representing in some 00:37:06.062 --> 00:37:09.962 form the discrete element or the elements of these sequences. 00:37:10.602 --> 00:37:19.142 So how do we know that this is the neural substrate of the inference we want to look at? Okay. 00:37:19.322 --> 00:37:26.062 One of the limits of the approach is that neurophysiology in non-human primates 00:37:26.062 --> 00:37:32.582 with the method of electrophysiology is unable to conclude about causality. 00:37:33.162 --> 00:37:41.322 So, the only possibility for a study exploring whatever behavioral control in 00:37:41.322 --> 00:37:48.202 whatever in whatever experimental setup is to find a relationship between, 00:37:49.807 --> 00:37:53.887 controlled variable and the measured variable. So it's a relationship. 00:37:54.207 --> 00:37:58.907 So in our case, we found a strong relationship between a behavioral phenomenon, 00:37:59.287 --> 00:38:05.467 two behavioral phenomena, that is the symbolic distance and the serial position effect. 00:38:09.327 --> 00:38:17.927 Being unable to find such a relation will allow us to conclude that the neurons code, 00:38:17.927 --> 00:38:24.067 code or the firing rate of neurons in prefrontal cortex is unable to participate 00:38:24.067 --> 00:38:33.187 properly to the behavioral counterpart of the transitive inference task. 00:38:33.647 --> 00:38:39.467 But we are unable, again, for the limitation of the method, to conclude that 00:38:39.467 --> 00:38:48.147 the prefrontal cortex is necessary and that the task is controlled selectively by that area. 00:38:48.227 --> 00:38:51.587 That actually is something that I don't believe and other people don't believe 00:38:51.587 --> 00:38:57.407 too in the area because we know that patients with the lesion at the level of 00:38:57.407 --> 00:39:03.567 the hippocampus or the prefrontal cortex or the parietal cortex has a different deficit, 00:39:03.687 --> 00:39:07.507 but always with the deficit that, 00:39:09.647 --> 00:39:16.887 imply that this subject are unable to solve a task like the transitive inference one. 00:39:17.087 --> 00:39:24.087 Okay. But now tell me if I look at the serial position versus symbolic distance, 00:39:24.247 --> 00:39:30.167 how is the neural response to the symbolic distance different from the serial position response? 00:39:32.727 --> 00:39:37.547 So if I would give you a neuron, a response profile of a neuron and I say well 00:39:37.547 --> 00:39:40.867 we measure this neuron while the monkey is performing this task would you be 00:39:40.867 --> 00:39:45.307 able just by looking at this response profile say whether it does the serial 00:39:45.307 --> 00:39:47.807 position or the symbolic relation? 00:39:48.307 --> 00:39:53.207 I think that what the analysis at the level of the population tell us is that 00:39:53.207 --> 00:40:03.147 every neuron is able to provide a contribution to the two factors but maybe 00:40:03.147 --> 00:40:06.127 be sometimes too noisy to reach statistical significance. 00:40:06.627 --> 00:40:14.967 So the first numbers that I provided is that with a strong statistical support 00:40:14.967 --> 00:40:17.587 there are neurons that are well coded for the, 00:40:19.007 --> 00:40:25.087 symbolic distance or they are well coded for the serial position because this 00:40:25.087 --> 00:40:30.147 doesn't need to have only one response different but also to follow the model. 00:40:30.307 --> 00:40:35.067 So you need to follow the model of the behavior And because the requirements 00:40:35.067 --> 00:40:37.927 are too many for a single neuron, it could be too much. 00:40:39.027 --> 00:40:42.987 So the answer is that there are neurons that are. 00:40:44.982 --> 00:40:48.242 More easy to it's easier to find neurons in 00:40:48.242 --> 00:40:51.282 prefrontal cortex that remember is an area that often 00:40:51.282 --> 00:40:54.802 doesn't provide in the discharge rate 00:40:54.802 --> 00:40:57.562 code so much contribution to say 00:40:57.562 --> 00:41:02.142 differently neurons in prefrontal cortex are not the same of neurons in motor 00:41:02.142 --> 00:41:07.442 premotor or sensory areas the amount of discharge of this nuance is very low 00:41:07.442 --> 00:41:13.082 often so this means that the signal is a It's a noisy signal by definition because 00:41:13.082 --> 00:41:16.382 if you discharge rate is 10 hertz, 00:41:16.542 --> 00:41:20.482 10 spike per second, then the possibility to replicate trial by trial, 00:41:20.562 --> 00:41:26.882 the same code is very difficult than when your neurons discharge 80, 100 hertz. 00:41:27.182 --> 00:41:33.042 But it was surprising that some of the neurons you reported are at a relatively elevated firing rate. 00:41:33.202 --> 00:41:38.502 Like you reported neurons up to 50 hertz, which seems rather than exceptional for that area. 00:41:40.482 --> 00:41:46.342 It's true, but I still consider up to 50 Hz a low rate. Okay. 00:41:47.002 --> 00:41:55.722 All right. So then, so okay, here we have- Anyway, in the delayed epoch is where 00:41:55.722 --> 00:41:58.342 these neurons are more able to express themselves. 00:41:58.542 --> 00:42:02.622 Generally speaking, the delay epoch is where the prefrontal cortex has been 00:42:02.622 --> 00:42:07.762 often studied for one of the characteristics that is assigned to the prefrontal 00:42:07.762 --> 00:42:09.202 cortex, which is the working memory. 00:42:09.322 --> 00:42:16.722 Right. So the working memory is often very well able to activate this news and the delay activity too. 00:42:17.182 --> 00:42:22.342 So now we have our transitive inference task, and you show monkeys can do it. 00:42:25.547 --> 00:42:29.187 Now, you have also taken this as a method 00:42:29.187 --> 00:42:34.027 if you want to look at certain neuropathologies, like schizophrenia. 00:42:34.507 --> 00:42:39.167 So, why is this relevant if we try to understand schizophrenia? 00:42:40.207 --> 00:42:45.107 Let's say that this is just the beginning of a story that we started recently. 00:42:45.107 --> 00:42:48.347 The reason is our 00:42:48.347 --> 00:42:55.787 goal is to link pharmacology and the electrophysiology with different methods 00:42:55.787 --> 00:43:03.807 in particular the idea of this study that I present today is that it's well 00:43:03.807 --> 00:43:07.247 known that schizophrenic patients have deficit, 00:43:08.687 --> 00:43:12.247 in solving the transitive inference task. 00:43:14.907 --> 00:43:22.987 Since there is a drug that is ketamine that has been used often in animal model to provide, 00:43:23.747 --> 00:43:29.507 one of the possible model of 00:43:29.507 --> 00:43:34.727 schizophrenia, that is the one related to the dopamine use in the brain. 00:43:35.827 --> 00:43:44.027 We tested if ketamine as a receptor agonist of the NMDA, that is one of the 00:43:44.027 --> 00:43:45.367 receptor of dopamine in the brain, 00:43:47.067 --> 00:43:56.687 it's able to modulate the behavioral response of the animal that we trained 00:43:56.687 --> 00:43:57.807 in the transitive inference. 00:43:57.807 --> 00:44:04.227 And the figure that comes out is that the answer is yes, the ketamine is able 00:44:04.227 --> 00:44:08.607 to reduce the efficiency of this animal to solve the task. 00:44:08.827 --> 00:44:16.647 What is interesting is that the performance is particularly reduced more in 00:44:16.647 --> 00:44:20.587 the figures that are considered transitive. 00:44:20.587 --> 00:44:30.727 That means those that are about comparison between never experience and the Falshermore, 00:44:30.827 --> 00:44:39.867 those that does not include the extreme spares that others refer as an anchor point. 00:44:41.013 --> 00:44:46.353 So, but what you see in the monkey was a degradation of performance? 00:44:46.953 --> 00:44:52.073 Yeah, generally it was quite similar for all the comparison, 00:44:52.233 --> 00:45:03.113 but we partially explored the data and the result is that the statistical significance 00:45:03.113 --> 00:45:10.493 is particular for the comparison that are more related to the transitive inference capability. 00:45:11.013 --> 00:45:17.633 Okay but then that's also exactly what was observed in schizophrenic patients 00:45:17.633 --> 00:45:24.593 so schizophrenic patients has no problem on concluding which one is has more value of A and B, 00:45:24.653 --> 00:45:30.953 B and C on A and F including the anchor point or B and F because once again 00:45:30.953 --> 00:45:40.473 it doesn't include the anchor point while when the question is which is higher between B and E E, 00:45:40.493 --> 00:45:43.213 then the deficit emerges, 00:45:43.433 --> 00:45:50.653 and the same happens in our monkey under sub-anesthetic doses of ketamine. 00:45:51.373 --> 00:45:54.553 But now, how do you explain that in terms of the action of ketamine? 00:45:56.433 --> 00:46:02.153 Okay, I think it's one of the possibilities, once again, to believe that the 00:46:02.153 --> 00:46:10.093 way of the representation is on the brain at the level of the neurons, it's very noisy. 00:46:10.233 --> 00:46:17.433 And that the ketamine reducing the efficiency of dopamine is increasing the 00:46:17.433 --> 00:46:19.993 noise and then making more difficult to the comparison. 00:46:20.513 --> 00:46:25.973 But now schizophrenia is also affecting, let's say, experience, 00:46:26.133 --> 00:46:27.753 subjective experience, right? 00:46:27.993 --> 00:46:30.493 So it has a very specific phenomenology as well. 00:46:30.913 --> 00:46:33.273 So do you believe that the same holds for the monkey? 00:46:35.117 --> 00:46:40.457 So in other words, is this task actually also informing us about the role of 00:46:40.457 --> 00:46:44.537 consciousness in decision-making, or you see it really as disconnected from those phenomena? 00:46:44.777 --> 00:46:47.677 I think it's very difficult to conclude such a... 00:46:49.337 --> 00:46:53.617 No, but I'm just... I mean, we have no evidence that the monkey has... 00:46:54.257 --> 00:46:57.497 We know that the ketamine is a dissociative drug. 00:46:57.657 --> 00:47:06.317 Yeah, exactly. But I don't know how much our dosage was able to really produce 00:47:06.317 --> 00:47:08.257 a monkey dissociated from the reality. 00:47:08.837 --> 00:47:14.457 Because we want to have the monkey still able to perform the task. 00:47:14.517 --> 00:47:19.117 Actually, the monkey was still performing the task with a good percentage of success. 00:47:20.037 --> 00:47:25.937 But the deficit was selective for a portion of the task, not for the whole task. 00:47:25.937 --> 00:47:33.617 So I don't know if this could be used as a very easy, I think, result, 00:47:33.997 --> 00:47:41.497 might be too simple to argument that maybe the monkey was consciously unable 00:47:41.497 --> 00:47:44.837 to participate to the transitive inference. 00:47:45.197 --> 00:47:47.937 That is a conclusion. No, that would not be my conclusion. 00:47:48.317 --> 00:47:52.997 My conclusion would be more like, if you use this paradigm as a way to also 00:47:52.997 --> 00:47:53.577 investigate schizophrenia. 00:47:54.857 --> 00:48:00.017 Is it then not by implication the case that we are implicitly also probing conscious 00:48:00.017 --> 00:48:03.957 states and conscious operations in this monkey, and not only logical ones? 00:48:05.597 --> 00:48:09.737 Yeah, it needs to be explored. Of course. No, it's clear. Yeah, yeah. 00:48:10.577 --> 00:48:16.337 Since the dragon is able to modulate consciousness, it's by definition a possibility. 00:48:16.337 --> 00:48:18.377 Yeah, exactly. I agree with you. 00:48:19.457 --> 00:48:26.337 So So, basically the conclusion then is that, okay, you also on top of that showed, 00:48:26.597 --> 00:48:33.657 which was very interesting, that if you now in this task go to a next level 00:48:33.657 --> 00:48:38.997 of description of the physiology of the monkey brain, which is the local field potential, 00:48:39.377 --> 00:48:45.237 that you also find correlates there of this transitive inference. 00:48:46.377 --> 00:48:53.217 So, what did you observe there exactly? Okay, for us the analysis of local field 00:48:53.217 --> 00:49:01.777 was a direct consequence of the observation in humans that fMRI is modulated by humans. 00:49:02.376 --> 00:49:08.256 By the transitive comparison in a transitive inference task. 00:49:08.716 --> 00:49:14.656 And since it's commonly accepted that the most close, 00:49:15.776 --> 00:49:21.196 neurophysiological signal to the bold activity of fMRI is the local field, 00:49:21.376 --> 00:49:27.056 for us it was very important to look also to the local field modulation during the task. 00:49:27.476 --> 00:49:33.436 So it probably was not a surprise anymore more after finding that the single 00:49:33.436 --> 00:49:38.456 neuron firing rate was modulated to find that also that the local field was modulated. 00:49:39.676 --> 00:49:52.036 But it's also important to show evidence of this because some way put the, again, 00:49:52.136 --> 00:49:55.496 the level of the multiscale, that means local computation, 00:49:55.616 --> 00:50:01.116 remote computation that need need to be further studied in the future because 00:50:01.116 --> 00:50:04.396 local field has access to the synaptic integration, 00:50:04.696 --> 00:50:11.416 so we don't know from where the signals came, and to local reverberation from the level of the gamma. 00:50:12.756 --> 00:50:19.536 Even if we haven't been able to show a strong effect at the level of the gamma in the local field, 00:50:19.616 --> 00:50:24.916 but at the level of the low components that are those more related to the input 00:50:24.916 --> 00:50:29.436 from From the thalamus or from the other cortical area like the hippocampus, 00:50:29.436 --> 00:50:31.396 once again, or the posterior parietal region. 00:50:31.756 --> 00:50:38.196 But you seemed to show that the deflections in this local field potential correlated 00:50:38.196 --> 00:50:41.036 very specifically with properties of the task. 00:50:41.256 --> 00:50:46.956 Absolutely. Actually, we have been able to find a very strong relationship more 00:50:46.956 --> 00:50:50.256 with the local field than with a single unit. Mm-hmm. 00:50:50.934 --> 00:50:59.494 And the reason could be different. One is that the single neuron expressed their 00:50:59.494 --> 00:51:02.054 contribution with different tuning forms. 00:51:02.954 --> 00:51:09.774 Sometimes with a shape very similar to the performance, like in the serial position, a U-shape. 00:51:10.034 --> 00:51:12.194 Other times it was inverted. 00:51:13.234 --> 00:51:20.474 And as I said before, Therefore, the noisy level of the single-neuron was sometimes 00:51:20.474 --> 00:51:22.094 too much to reach significance. 00:51:23.794 --> 00:51:32.794 And also the time in the task from the pair presentation to the decision time, 00:51:32.934 --> 00:51:36.334 that is after the go signal during a delay, 00:51:37.094 --> 00:51:43.314 for the single-neuron was never, I'll say, often not the same. 00:51:43.314 --> 00:51:47.354 I mean that the single neurons are able to express their contribution at the 00:51:47.354 --> 00:51:51.734 beginning of the delay, at the middle part of the delay, or the late portion of the delay. 00:51:51.914 --> 00:51:57.434 While for the local field, it was very time-located to the pair presentation 00:51:57.434 --> 00:51:59.114 at about 400 milliseconds. 00:51:59.694 --> 00:52:06.494 And this allowed to extract probably a better signal to find a correlation as the one we found. 00:52:07.334 --> 00:52:11.014 But it's interesting to see that you don't have such a tight correlation with 00:52:11.014 --> 00:52:15.894 your single-cell recording. So would that suggest that there's another dynamic 00:52:15.894 --> 00:52:18.694 process at work that shapes this local field potential? 00:52:19.654 --> 00:52:25.214 Like other synaptic activities that you don't pick up with your single cell 00:52:25.214 --> 00:52:27.314 measurements that shape the local 00:52:27.314 --> 00:52:34.254 field potential? Yeah, the single unit activity is really local, okay? 00:52:34.394 --> 00:52:41.774 While the local field, by definition, is able to sample at least a volume that 00:52:41.774 --> 00:52:45.054 is huge compared to a single unit. 00:52:45.834 --> 00:52:52.114 So, for sure, the conclusion is that we are not looking at the same computational level. 00:52:52.274 --> 00:52:59.654 Right. and actually the better performance of the single unit coding is when 00:52:59.654 --> 00:53:02.894 the neurons are considered together forming a population. 00:53:04.314 --> 00:53:12.994 Local field is by definition a mesoscopic approach so this is probably expected 00:53:12.994 --> 00:53:20.914 that if a signal is there it's much easier to be found and maybe correlated with behavior so So, 00:53:20.914 --> 00:53:26.314 the experiments you described basically suggest that the monkey that we investigate, 00:53:26.534 --> 00:53:32.294 and possibly also humans, organize the information that they need in this transitive 00:53:32.294 --> 00:53:38.074 inference task on a line of magnitude, of magnitude relations in some sense. 00:53:38.314 --> 00:53:43.174 And this line is very well demarcated at the beginning and the end. Mm-hmm. Okay? 00:53:43.474 --> 00:53:49.714 Correct. So, how long can those sequences be, you think, for the monkey? Okay. 00:53:50.547 --> 00:53:55.807 For our monkey, I think they need to vanish soon. 00:53:56.347 --> 00:54:03.307 However, I haven't shown this data today, but we asked the monkey some time 00:54:03.307 --> 00:54:10.007 to play with the series that was learned the day before, and the series is still there. 00:54:10.247 --> 00:54:13.747 Even if the monkey knows that it's better 00:54:13.747 --> 00:54:28.007 for him to have a very good performance to forget completely what was the assignment 00:54:28.007 --> 00:54:31.587 of figures the day before because they are never used. 00:54:31.827 --> 00:54:37.247 Right, exactly. But once again, there is something that's implicit probably. 00:54:38.667 --> 00:54:45.427 There's something there. I'm not thinking to use them, but if I'm asked to use them, I'm ready to. 00:54:45.507 --> 00:54:51.927 The performance was very rapid and perfect the day after. 00:54:52.247 --> 00:54:57.947 We have no tested other days, I mean two days or three days, 00:54:57.947 --> 00:55:01.847 so I don't know how long it remains. It's pretty astonishing, yeah. 00:55:02.087 --> 00:55:05.927 But then the other thing is, so there's this famous, or the snark effect that 00:55:05.927 --> 00:55:09.067 Stan De Haan has written a lot about and others, 00:55:10.567 --> 00:55:15.767 which shows that also humans, when they have to deal with magnitude judgments, 00:55:16.107 --> 00:55:19.187 let's say something is bigger or smaller than something else and so on, 00:55:19.307 --> 00:55:21.327 that there's a very specific spatial bias. 00:55:21.607 --> 00:55:26.887 So if you have to respond with your left hand, you respond more quickly to small, 00:55:27.127 --> 00:55:31.907 lower magnitudes, and with the right hand, you're faster. when the magnitude is higher. 00:55:32.047 --> 00:55:37.027 Also suggesting that there is some spatial organization of magnitude judgments. 00:55:37.187 --> 00:55:41.567 And what I found interesting about that is if we compare that to your results, 00:55:42.307 --> 00:55:48.127 you interpret your results as also the title of your talk suggested as the ability 00:55:48.127 --> 00:55:49.647 to perform logical operations. 00:55:51.210 --> 00:55:56.890 But in some sense, I could also argue, well, look, maybe this is all just spatially organized. 00:55:56.990 --> 00:56:01.090 And I use associative rules, I use analog reasoning, if you want, 00:56:01.250 --> 00:56:03.590 to then make these magnitude judgments. 00:56:03.790 --> 00:56:08.490 So, is it really necessary to interpret your results in terms of logical operations? 00:56:09.370 --> 00:56:14.190 It's just the definition of logic. Maybe you as a psychologist, 00:56:14.450 --> 00:56:22.970 if I'm not wrong, by training your definition of logic is much more stronger 00:56:22.970 --> 00:56:25.190 than the one that I'm using at the moment. 00:56:27.730 --> 00:56:39.830 And I agree that the monkey is using a space as a model to solve a task. 00:56:39.830 --> 00:56:46.070 If the solution of this task is logic it's a term of definition and in my point 00:56:46.070 --> 00:56:53.730 it's logic whatever is more than what was available by by, 00:56:55.010 --> 00:57:02.270 the information provided at the beginning so it's not a level of logic that, 00:57:03.090 --> 00:57:09.310 is probably in your mind but it's a first level of be able to, 00:57:10.236 --> 00:57:13.776 to conclude if then right 00:57:13.776 --> 00:57:16.956 it's very elementary but remember 00:57:16.956 --> 00:57:23.596 that these are non-human primates yes and and as an animal we don't know if 00:57:23.596 --> 00:57:27.696 they are able to we have no argument to conclude that they are in particular 00:57:27.696 --> 00:57:36.756 for non-verbal conclusion able to to have a to use arguments properly 00:57:37.076 --> 00:57:38.756 to obtain new information. 00:57:39.816 --> 00:57:43.956 But it also means that maybe what you're showing us here is an alternative way 00:57:43.956 --> 00:57:49.616 how we can think about cognitive operations that give rise to behavior regularities 00:57:49.616 --> 00:57:52.616 that we as observers can interpret in logical terms, 00:57:52.756 --> 00:57:55.276 but it actually follow very different principles. 00:57:56.136 --> 00:58:00.876 And the same might hold for our own cognitive operations, right? 00:58:00.876 --> 00:58:08.236 That we also then, let's say now, a posteriori, describe them as logical operations. 00:58:08.716 --> 00:58:12.436 But actually, in terms of the internal operations, it's much more an analog, 00:58:12.616 --> 00:58:13.976 spatially organized process. 00:58:14.436 --> 00:58:18.416 I mean, would you consider that still as an open option? I agree with this. 00:58:18.736 --> 00:58:24.336 Actually, it's a behavioral experiment that we perform in humans, 00:58:24.596 --> 00:58:26.836 but I think it's important to describe. 00:58:26.836 --> 00:58:31.456 At this point, at the beginning of our testing with animals, 00:58:31.656 --> 00:58:39.196 we asked normal subjects at the student's level of age to perform a transitive 00:58:39.196 --> 00:58:42.336 inference task while changing their gates. 00:58:42.756 --> 00:58:49.176 So, looking at the left or to the right. So, for the visual attributes of the 00:58:49.176 --> 00:58:55.676 symbols and the primary goal of the subject was not related at all with the gates. 00:58:55.676 --> 00:59:00.236 But still the gauge is able to interfere as ketamine. 00:59:01.080 --> 00:59:04.680 With the subject performance okay so 00:59:04.680 --> 00:59:12.060 looking left and right in a whatever it's like changing a reference frame for 00:59:12.060 --> 00:59:15.520 a mental operation right exactly it's not a logical operation maybe i agree 00:59:15.520 --> 00:59:21.180 with you but it's a mental operation it determines that in in a space that is no physical, 00:59:21.880 --> 00:59:29.480 these symbols need to be represented right and this space could be uh modulated by a physical. 00:59:31.220 --> 00:59:37.800 Signal that is the angle of gaze that normally is well known to be able to modulate 00:59:37.800 --> 00:59:42.220 visual response in the space. Exactly right. 00:59:43.220 --> 00:59:48.060 We're saying the same thing. It's also on that account that I find that your 00:59:48.060 --> 00:59:49.700 results I think are very exciting. 00:59:49.980 --> 00:59:55.380 But now, so you're active in systems neuroscience, you look at these really 00:59:55.380 --> 01:00:01.900 complex processes, logical operations in monkeys, whether They are intrinsically logical or not. 01:00:02.200 --> 01:00:04.240 It's very interesting and very exciting. 01:00:04.760 --> 01:00:07.940 You started as a neurologist, switching over to neuroscience. 01:00:08.380 --> 01:00:15.560 So if we would like to follow the study of the brain in the tradition of Stefano 01:00:15.560 --> 01:00:18.640 Ferreira, what would be the Stefano law of the study of the brain? 01:00:19.440 --> 01:00:27.920 I think that is a little bit far from some rules that are currently available, 01:00:28.760 --> 01:00:31.060 at the financing level in Europe. Okay. 01:00:31.260 --> 01:00:36.240 I want first to know exactly how the brain works and then maybe to model it, 01:00:36.380 --> 01:00:42.480 to understand properly and to translational apply to this knowledge to pathology. Okay. 01:00:43.400 --> 01:00:48.500 So and then predictions. So soon I'm going to come to Rome four years from now. 01:00:48.560 --> 01:00:49.600 I'm going to go to your lab. 01:00:50.080 --> 01:00:53.060 I'm going to ask you, okay, four years ago you gave me this prediction. 01:00:54.380 --> 01:00:58.460 Show me whether you tested it and what was the outcome. So what's the one prediction 01:00:58.460 --> 01:01:02.340 you would make today that you feel most committed to? 01:01:03.908 --> 01:01:07.368 For the four years? Next four years? Yeah, four years, yes. Okay, 01:01:07.528 --> 01:01:13.048 let's consider first that monkey experiments, primate experiments in general are very slow. 01:01:13.248 --> 01:01:17.528 So in four years I will be able probably to complete just one experiment. 01:01:18.108 --> 01:01:24.328 Or let's say to have one finished at the second at the beginning of the trip. 01:01:25.028 --> 01:01:35.868 But the next for me will be to be able to record from from many areas of the brain simultaneously. 01:01:37.708 --> 01:01:43.608 In this task and other tasks that I'm using at the moment in the lab, 01:01:43.708 --> 01:01:44.928 we are using it in the lab, 01:01:46.348 --> 01:01:51.508 and at the same time to have assessed to different level of the neural signal, 01:01:52.348 --> 01:01:59.788 to provide a better decode to whatever is going on in the population of neurons 01:01:59.788 --> 01:02:02.688 controlling the behavior that I'm interested. 01:02:04.068 --> 01:02:14.068 So in particular for the transitive inference task we are already studying the 01:02:14.068 --> 01:02:21.288 premotor cortex and the goal is to go to the posterior parietal following what is the, 01:02:22.368 --> 01:02:29.008 network described in humans and by having access to a high resolution signal 01:02:29.008 --> 01:02:34.088 to provide details on what is still unknown, 01:02:34.308 --> 01:02:40.948 on the possibility of the brain to provide support to such a behavior. 01:02:42.948 --> 01:02:47.488 But that seems very methodological. Do you have a specific prediction you're testing in doing this? 01:02:50.028 --> 01:03:04.408 The prediction is that it's not very singular, but I still believe that need to be confirmed. 01:03:07.188 --> 01:03:14.968 Properly that the behavior like the one that I'm studying need to be coordinated by different areas. 01:03:14.968 --> 01:03:22.268 I would like to find the key of how this coordination happens. 01:03:22.688 --> 01:03:26.528 Okay, great. Stefano Ferreira, thank you very much for this conversation. Thank you, Paul. 01:03:30.788 --> 01:03:36.468 The CSN podcast was produced by the Convergent Science Network of Biometrics 01:03:36.468 --> 01:03:42.908 and Biohybrid Systems, a project funded by the European Sevens Research Framework Program. 01:03:44.968 --> 01:03:49.728 For more interviews, recorded lectures, or upcoming conferences in the field 01:03:49.728 --> 01:03:55.748 of biometrics and biohybrid systems, go to csnnetwork.com. 01:03:56.080 --> 01:04:05.040 Music.