Archive for the ‘modularity’ Category

How are values computed in the brain? Rewards can be as many things: the expectation when having just ordered your favourite dish; the child’s joy at Christmas Eve; the enjoyment of good music or the wonderful taste of strawberries.

But how does the brain process these many different kinds of rewards? Does it treat all types of rewards equally or does the brain distinguish between different kinds of rewards? Rewards can come in many different forms: from sex, social recognition, food when you’re hungry, or money. But it is still an open question whether the brain processes such rewards in different ways, or whether there is a “common currency” in the brain for all types of rewards.

Guillaume Sescousse and his colleagues in Lyon recently reported a study on how the brain reacts differently to money and sex. A group of men were scanned with functional MRI. While being tested, subjects played a game in which they sometimes reveiwed a reward. The reward could be money or it could be the sight of a lightly dressed woman. So there were two types of rewards. Money can be said to be an indirect reward, and the sexual images can be seen as more immediately rewarding (at least for most heterosexual men). But how did the brain process these rewards?

The researchers found that there were unique activations for both sex and money, but that there were also overlapping regions of activity. On one hand, for both types of reward was a general activation of what we often refer to as the brain’s reward system (ventral striatum, anterior insula, anterior cingulate cortex and midbrain; see figure 1). The brain thus uses the some structures to respond to both types of reward.

Regions of common activations

But there were also specific activations for erotic pictures and money. And this difference was primarily made in the brain’s prefrontal cortex, especially the orbitofrontal cortex (OfC). Here, it was found that monetary rewards engaged more anterior OfC regions, while erotic images activated more posterior OfC regions.

This could suggest that the brain also treats the two types of reward differently. The crux of this paper, however, is how one explains the difference. As noted, the researchers used two different kinds of reward, but they differ in several ways which I will try to summarize here:

  • Direct vs indirect
    • Money is indirectly rewarding, because money can not be ‘consumed’ in itself. They are rewarding to the extent they could be exchanged for other things. Erotic images are in themselves directly rewarding. Not because they symbolize sex, or the possibility of sex, but because they have an immediate rewarding effects.
  • Abstraction level
    • Another option is to say that erotic pictures and money differ in their level of abstraction: Erotic images are concrete, while money is an abstract reward.
  • Time interval
    • A final possibility is that there are differences in the time interval: Erotic images are immediately rewarding, while the money can only be converted into real value after a while (for example, after scanning, or after a few days where you spend the money). We already know that the frontopolar regions of the brain is among the regions that are most developed in humans compared to other primates, and is linked to our unique ability to think about the future, i.e. prospective memory and planning, and through this to use complex abstractions for rewards, including money.

Regions of distinct activations: orange = monetary rewards, green = sexual rewards

What the exact cause of this common currency as well as the separation between money and erotic pictures is still unclear and warrants further studies (which I am currently undertaking). The essential addition of this study is the separation between the posterior and anterior parts of the OFC in processing different kinds of rewards. By showing common and distinct regions, this study may resolve some of the ongoing debates in the decision neuroscience / neuroeconomic literature. But as always found in science, this study generates more questions than it resolves, and we can only hope that future studies can add to this knowledge.


Read Full Post »

aniston.jpegHow specific — or sparse — is the neural representation of a memory trace? Quian Quiroga and colleagues now have an article in Neuron (PDF), where they describe their well-known studies using single-cell recordings to well-known faces. As you most likely know, this has given rise to the debate about the “Jennifer Aniston neuron”. Their findings, briefly put, have demonstrated that single cells show quite specific responses to very specific visual stimuli. While one cell may have a preferential response to the Sidney Opera House, another responds dramatically more to Hale Berry, while yet another cell responds to, well, Jennifer Aniston.

The Quiroga studies have re-iterated the debate (if it ever went dead) about how specific the neural coding is in the brain. Is it really so that the brain has such a specific code that one cell can represent one percept? Do we have a grandmother cell, a President Nixon cell and Marilyn Monroe cell?

Today, there is wide agreement that the one-cell-one-percept idea is untenable and unsupported by the literature. Rather than cingle cells, we see that networks represent a percept, rather than single cells. However, the findings by Quiroga et al. have nevertheless stunned the scientific (and global) community with regard to just how specific the neural code can be, and that it can be detected in a single neuron. The findings that we can record how one single neuron responds to one, and only one, percept, is quite surprising.

So forget about the grandmother cell, right? Or maybe not. After all, following the idea from these findings, we should not be surprised that there would in fact be one neuron that responded preferentially to our grandmother. Yes, it would be an expression of a “network code” representing our grandmothers as such, but nevertheless, you may in fact have that one neuron that responds to ol’ granny.

While we leave it at that, it is still surprising that this team of researchers use the term “medial temporal lobe” (or MTL for short). Why doe they say that there is sparse coding in the MTL? It’s a rather big region, and a region packed with qualitatively different regions. Not only are these regions different anatomically, but also functionally, they are thought to be involved in different functions. The perirhinal cortex is involved in processing (and encoding) of complex visual objects as well as novelty processing and working memory, the (posterior) parahippocampal cortex is involved in spatial processing (remember the parahippocampal place area). The entorhinal cortex has a medial and lateral part that deal with spatial and object information, respectively. And in addition to the hippocampus and amygdala, with their quite different functions, we may extend the MTL concept to include the temporopolar cortex, and maybe even the inferotemporal cortex.

Where in this complex system do Quiroga et al. find their sparse coding? Everywhere? My bet is on the perirhinal cortex due to its involvement in complex visual object processing. I’ve come to know that the researchers have not had structural scans available to determine the exact location of their electrodes. The scans have obviously been made, but they have not been able to use that information (hush hush, don’t tell anyone…).

So, while these studies are indeed important to our understanding of the coding of specific information, we’re left with a huge gap in terms of their anatomical properties. While most of the research community focuses on MTL subdivisions to an increasing extent, it is a bit puzzling to me that nobody have ever criticised these studies for their sparsity of anatomical information. Maybe I’ll be the first?


Read Full Post »

broca.jpgNature is running a nice news article on the re-localization of Broca’s language area in the brain, and has as feature about it in their latest podcast.

Pierre Paul Broca originally described patient cases in which the patient suffered speech production deficits following injury to the left frontal hemisphere. However, a revisit to Broca’s original papers (see translations here and here), combined with a modern scanning of the preserved remains of Broca’s patients, has revealed that what has been called Broca’s area in modern times does not correspond to the areas implicated by Broca in his patient descriptions and neuroanatomical descriptions.

The story is interesting, but I’m amazed that the excitement is running so high. After all, lots of papers have already dethroned Broca’s (and Wernicke’s) area in the role of language processing. Take the example of the special issue of Cognition on language. Basically, what we know about language in the brain is beyond the talk (!) about Broca and Wernicke, and especially the models they suggested. Rather, both language comprehension and production require a larger neural symphony, and with substantial internal redundancy. IOW, Broca’s area can participate in comprehension, and Wernicke can play a part in production.

Nevertheless, the Nature news article is a good read, and I always recommend the Nature podcast.


Read Full Post »

figure1.jpgIs binding the single most important concept in neuroscience? I think it is, even without making the concept too general or vague. On the contrary, binding seems to be a general concept to understand the workings of the brain. No more need for modules of perception, cognition, memory and action. Binding is the solution.

More specifically, what is binding? Or, to reframe the question 100%: what happens when the brain works? To many, the brain binds information together at all levels throughout the brain. If you perceive an object, that particular object is a mixture between colour, form, position, movement etc., that is bound together. Because of you look at the early sensory processes in the brain, we know that the features of an object are treated by separate processes in the brain. Accordingly, they can be lesioned separately, leading to e.g. acquired colour blindless but with intact movement perception.


Read Full Post »

In an interesting paper in the latest version of Progress in Neurobiology, Yuri I. Arshavsky from UCSD writes about the epistemological dualism that exists in modern neuroscience. basically, Arshavsky claims that there is a covert dualism in the way that neuroscientists are treating mind-related topics, especially the study of “consciousness”. Indeed, as he claims:

This covert dualism seems to be rooted in the main paradigm of neuroscience that suggests that cognitive functions, such as language production and comprehension, face recognition, declarative memory, emotions, etc., are performed by neural networks consisting of simple elements.

This might initially sound a bit strange. Is not cognitive functions such as face perception due to operational simple elements? Face perception as such is a combination of many simple processes that operate in unison. So what is Arshavsky proposing? Indeed he suggests the existence of a certain kind of brain cells:

(The) performance of cognitive functions is based on complex cooperative activity of “complex” neurons that are carriers of “elementary cognition.” The uniqueness of human cognitive functions, which has a genetic basis, is determined by the specificity of genes expressed by these “complex” neurons. The main goal of the review is to show that the identification of the genes implicated in cognitive functions and the understanding of a functional role of their products is a possible way to overcome covert dualism in neuroscience.

So there should exist a subset of neurons that integrate information from a variety of input. This sounds strange, since all neurons integrate inputs from thousands of inputs, many from a large variety of inputs. So what are complex neurons? Here, we are told that:

(…) neural networks involved in performing cognitive functions are formed not by simple neurons whose function is limited to the generation of electrical potentials and transmission of signals to other neurons, but by complex neurons that can be regarded as carriers of “elementary” cognition. The performance of cognitive functions is based on the cooperative activity of this type of complex neurons.

In this way, complex neurons seem to be integrative neurons, i.e. cells that integrate information from a variety of processes. This could include the multi-modal neurons found in the functional sub-structures of the medial temporal lobe, such as the hippocampus, perirhinal, entorhinal and temporopolar cortex. But would it not mean the colour processing nodes in the visual cortex? Which IMO leads us back to a basic question: what is a functional unit in the brain. yes, the neuron is a basic building block of information processing in the brain. But what is special about language, memory and so forth in the brain?

It is possible that Arshavsky is not radical enough: what we should seek out is to avoid using generalistic and folk-psychological concepts in the first place. We should possibly not study “language”, “memory” or “consciousness”, since these concepts will always allude to fundamental assumptions of “language-ness”, “memory-ness” and “consciousness-ness”, IOW that there is something more to explain after we have found out how the brain produces what we recognize and label a cognitive function.

Maybe neuroscientists are not using a poor strategy after all? Maybe ignoring the past history of philosophy of mind is the best solution. I’m not sure (nor am I sure that I represent Arshavsky’s view properly). But how we choose to label a cognitive function depend on our past historical influence and learning, as well as our current approach.


Read Full Post »

mtl.jpgTake any textbook on cognitive neuroscience. If you go through the book you wil see that there are chapters on perception (e.g. vision), memory, and language. Each chapter has its own vocabulary, theories and experimental evidence. Each chapter may even have been written by different authors (i.e. authorities).

Once reading such a book you will have knowledge about how visual input is processed from the initial steps in the retina, through the thalamic nuclei, and in the visual cortex, just as well as you will learn that as you perceive something as an object you will make use of areas in the temporal lobe, including the fusiform gyrus. You will have learned that memory — especially episodic and semanic memory — is a result of activity occurring in the medial temporal lobe, and especially hippocampus. You will know that theories of language and semantics point to the temporal lobe as important for its functioning.

All in all, you have a nice impression of how the brain is responsible for different perceptual and cognitive functions. But think now of the three examples: they all seem to imply the temporal lobe as important for their functioning. So does this mean that visual perception, memory and language resides in different, non-overlapping parts of the temporal lobe? If so, how do these areas or modules communicate with each other?. What is the lingua franca of neurons comunicating information fro the visual senses to memory and semantics? Add on top of this that parts of the temporal lobe has been implemented in many other functions, including hearing (e.g. the planum temporale and Heschl’s gyrus) and odour processing (e.g. the entorhinal cortex). How does this combine with the other functions? Should we see the temporal lobe as a patchwork of distinct and neatly segregated functions?

For a long time the predominating view of the temporal lobe has been a strictly modular one: one part of the lobe processes visual input, there are language and memory modules. Non-overlapping parts of a lobe that are tuned to process one kind – but not other kinds – of information.

But this view is changing dramatically. Today, following researchers such as Elisabeth Murray, David Gaffan and others (especially from the universities of Cambridge and Oxford, UK) the standard view of temporal lobe function is changing. Instead of a functionally segregated model of the temporal lobe, these researchers now suggest that the lobe has an entirely different way of functioning. In this area, often referred to as the medial temporal lobe one has now documented not only multiple cognitive functions in a brain area once thought to be dedicated to memory, but also redundancy between the structures. Some examples:

  1. There is a functional specialization within the rhinal cortices beyond the involvement in memory: the entorhinal cortex is involved in odour perception as well as multi-modal conjunct perception, i.e. the perception of the entirety of a scene, including sights, sounds and more. The perirhinal cortex is involved in novelty processing, higher-order visual conjunct perception and discrimination, as well as high-specificity semantic processing.
  2. Specific and small anatomical regions are involved in different cognitive functions. For example, the perirhinal cortex has been shown to be involved in memory processes (particularly visual object encoding, but also other forms), novelty processing, semantic processing and high-order visual perception and discrimination

While point 1 does not conflict with a modular view of the brain-mind, point 2 poses a serious problem to any modularist view of the human mind and brain. In many respects, findings now converge on a view of the brain that stresses functional redundancy and degeneracy. In other words: A) one structure can participate in many different functions; and B) many structures are necessary parts of any given cognitive function. A mapping of a 1:1 relationship between a cognitive function and its wetware is thus unsupported by today’s knowledge.

So take that cog-neurosci textbook again, scroll through its pages and ask yourself: how are these cognitive functions connected? Better still, take the chapter to your supervisor, lecturer or whoever you want and ask: “how does the temporal lobe deal with memory, language, visual perception and other multi-modal operations, and how are these processes tied together?” It would be interesting to hear the replies you get.


Read Full Post »

Recently Thomas wrote about a paper by Yulia Kovas and Robert Plomin in the May issue of TICS discussing the implications of the fact that a great number of genes – dubbed “generalist” genes – affect not one, but most cognitive abilities. One obvious implication is that, if most genes being expressed in the brain affect several areas of the brain, the massive modularity hypothesis (MMH) might not hold true. As Kovas and Plomin wrote in the conclusion to their paper:

Our opinion outlined in this article is that the generalist genes hypothesis is correct and that genetic input into brain structure and function is general (distributed) not specific (modular). The key genetic concepts of pleiotropy and polygenicity increase the plausibility of this opinion. Generalist genes have far-reaching implications for cognitive neuroscience because their pleiotropic and polygenic effects perfuse the transcriptome, the proteome and the brain. This is more than a ‘life-is-complicated’ message. DNA and RNA microarrays provide powerful tools that will ultimately make it possible for cognitive neuroscience to incorporate the trait-specific genome and transcriptome even if hundreds of genes affect individual differences in a particular brain or cognitive trait. The more immediate impact of generalist genes will be to change the way in which we think about the relationship among the genome, the transcriptome and the ‘phenome’ of the brain and cognition.

As Thomas was quick to remark, this idea is of course sure to infuriate proponents of the MMH. Therefore, it comes as no surprise that Gary Marcus and Hugh Rabagliati has a letter in next month’s TICS criticizing Kovas and Plomin’s article. Here is their argument for upholding the MMH:

Genes are in essence instructions for fabricating biological structure. In the construction of a house, one finds both some repeated motifs and some specializations for particular rooms. Every room has doors, electrical wiring, insulation and walls built upon a frame of wooden studs. However, the washroom and kitchen vary in the particulars of how they use plumbing array fixtures, and only a garage is likely to be equipped with electric doors (using a novel combination of electrical wiring and ‘doorness’). Constructing a home requires both domain-general and domain-specific techniques. The specialization of a given room principally derives from the ways in which high-level directives guide the precise implementation of low-level domain-general techniques. When it comes to neural function, the real question is how ‘generalist genes’ fit into the larger picture. Continuing the analogy, one might ask whether different ‘rooms’ of the brain are all built according to exactly the same plan, or whether they differ in important ways, while depending on common infrastructure. Kovas and Plomin presume that the sheer preponderance of domain-general genes implies a single common blueprint for the mind, but it is possible that the generalist genes are responsible only for infrastructure (e.g. the construction of receptors, neurotransmitters, dendritic spines, synaptic vesicles and axonal filaments), with a smaller number of specialist genes supervising in a way that still yields a substantial amount of modular structure.

The interesting thing about this discussion between Plomin and Marcus is the fact that the question that they raise can be investigated empirically, as Kovas and Plomin note in a reply to Marcus and Rabagliati:

Finding high genetic correlations means that genes must be generalists at the psychometric level at which
these traits have been assessed. Therefore, a genetic polymorphism that is associated with individual differences in a particular cognitive ability will also be associated with other abilities. The question is how these generalist genes work in the brain. Does a genetic polymorphism affect just one brain structure or function, which then affects many cognitive processes, as suggested by a modular view of brain structure and function (mechanism 1 in [Kovas and Plomin’s original article])? This model assumes that brain structures and functions are not genetically correlated – genetic correlations arise only at the level of cognition. Another possibility, which we think is more probable, is that the origin of the general effect of a genetic polymorphism is in the brain because the polymorphism affects many brain structures and functions (mechanisms 2 and 3 in [Kovas and Plomin’s original article]). Of course, some polymorphisms might have general effects via mechanism 1 and other polymorphisms might have general effects via mechanisms 2 and 3, as Marcus and Rabagliati suggest. Fortunately, this is an empirical issue about DNA polymorphisms that does not require resorting to metaphors such as house-building. We did not say that the case for mechanism 3 was proven, which is what Marcus and Rabagliati imply with their partial quote. The full quote from our article is: ‘In our opinion, these two key genetic concepts of pleiotropy and polygenicity suggest that the genetic input into brain structure and function is general not modular’. Pleiotropy (in which a gene affects many traits) is a general rule of genetics. Polygenicity (in which many genes affect a trait) is becoming another rule of genetics for complex traits and common disorders. As we point out, polygenicity greatly multiplies and magnifies the pleiotropic effects of generalist genes. A more empirical reason for suggesting that the origin of generalist genes is in the brain is that gene-expression maps of the brain generally indicate widespread expression of cognition related genes throughout the brain.

I second that sentiment. It would be a big step forward if the massively modularity discussion would move beyond mere speculation and become grounded in empirical data.


Kovas, Y. & Plomin, R. (2006): Generalist genes: implications for the cognitive sciences. Trends in Cognitive Science 10: 198-203.

Marcus, G. & Rabagliata, H (2006): Genes and domain specificity. Trends in Cognitive Science, in press.

Kovas, Y. & Plomin, R. (2006): Response to Marcus and Rabagliata. Trends in Cognitive Science, in press.


Read Full Post »

Older Posts »