What’s really striking about this study is that it’s arguably the best example to date (or at least, the best example that I know of–I don’t follow this literature closely) of the power that new structural MRI techniques provide to assess in vivo brain connectivity in humans. In this case, the authors used diffusion spectrum imaging, a technique that lets the researcher construct whole-brain images of white matter fiber density and then (using some sophisticated post-processing) plot the trajectories of those tracts. The authors defined a connection between regions as the presence of at least one fiber with end-points in both regions (the more terminating fibers, the stronger the connection). Given an N x N matrix (where N = 998 different brain regions in this case!) of connectivity strengths between regions, they could then apply the suite of network topology metrics to produce those pretty, pretty figures.

Lest you think this all sounds like black magic (as I suspect a reviewer or two did), Hagmann et al. provide evidence that these structure-based connectivity maps (a) are reliable across hemispheres and scanning sessions; (b) degrade gracefully in the presence of noise; (c) conform nicely to connectivity data obtained from more conventional anatomical tract tracing techniques in monkeys; and (d) are quantitatively very similar to maps obtained using functional resting-state data in the same participants.  The sheer breadth of analysis in this paper is really quite striking, and you’d have to nit-pick to find faults with the methodology.

That said, there’s one critical question that these results don’t really address, and that’s what the findings mean from a functional standpoint. it’s easy to make the general argument that a small-world network structure is A Good Thing ™ for the brain to have; but the (arguably) more interesting question is why the hubs are located in these particular brain regions. The fact that a majority of the hubs (including posterior cingulate, precuneus, lateral parietal cortex, and superior temporal sulcus) are components of the brain’s “default” or task-negative network is clearly no coincidence. So what functional purpose does this pattern of connectivity serve? Why do those regions that are maximally activated at rest have the broadest pattern of connectivity with the rest of the cortex? Or is it perhaps the other way around, so that these regions develop their default status precisely because they receive inputs from multiple sources, and are ideally situated to mediate transitions between different task sets? Clearly, many questions remain to be addressed (warning: a horribly cliched ending to this post is imminent), but the Hagmann et al. paper will probably turn out to be a pretty important piece of the puzzle (see, I warned you).

Hat-tip: Neurophilosophy.

In brief, Papassotiropoulos and colleagues conducted a whole-genome scan on 500,000 distinct single nucleotide polymorphism (SNPs) in a large Swiss sample, and identified a correlation with episodic memory performance in two genes. They subsequently replicated the association between one of the genes (Kibra) and memory performance in two independent samples. What’s striking isn’t just the presence of two successful replications (almost unheard of in a paper that’s first to identify a gene-behavior relationship—many effects of this sort fail to replicate at all in subsequent studies), but also the size of the effect. In the initial sample, T allele non-carriers (i.e., subjects who had 2 C alleles of the Kibra SNP) performed 24% better on a free recall task after a 5 minute delay. You often hear people write off the molecular genetic approach to studying cognitive differences on the grounds that individual genes account for only a fraction of the variance and it’d take dozens or hundreds of genes to form a meaningful account. What the Kibra effect and other similar studies suggest is that, at least for some traits, a handful of genes may actually account for a considerable portion of the variance.

While the discovery and replication of the Kibra-episodic memory association alone would be a high-impact finding, Papassotiropoulos et al. didn’t stop there. They then went on to conduct brain imaging analyses in both humans and mice, demonstrating that a truncated version of Kibra is densely expressed in the medial temporal lobe (a region heavily implicated in episodic memory formation), but not in other areas such as the frontal lobes. They suggest that Kibra may exert its effect on memory via modulation of hippocampal function, though the precise locus and mechanism of effect is currently unknown.

But wait! There’s more! The authors then went on to conduct an fMRI study, in which they imaged 15 T allele carriers and 15 non-carriers during performance of an encoding task (a face-profession association task). They observed selective increases in the medial temporal lobes in non-carriers (the group with poorer performance in the genetic samples) relative to carriers, and no regions showing the converse effect. Because the two groups were matched for delayed memory performance, the relative increase in the group with poorer performance likely reflects less efficient processing, requiring greater activation to achieve the same level of memory performance.

As if all this wasn’t enough, Papassotiropolous et al. also conducted structural imaging analyses using both automated whole-brain and manual tracing approaches. These analyses didn’t turn up any significant findings, but given the amount of effort and breadth of expertise required for all of these analyses, one can only applaud them for trying.

On the whole, I can’t say enough good things about this paper. Regardless of the implications of the substantive finding itself (which, if replicated by other groups, should have important implications both theoretically and practically), it’s remarkable to see such a diversity of approaches and sources of expertise brought to bear on a single problem. Lots of people pay lip service to the notion that inter-disciplinary science is a good thing, but to date there are relatively few demonstrations of the idea on a large scale. The combination of molecular genetics and cognitive neuroimaging seems like a particularly profitable approach, yet few people have applied it thus far (with several notable exceptions, e.g., centers at the NIH and Pittsburgh). If this is the shape of things to come, it’ll be fun to watch over the next few years as this sort of research takes off…

V1 is the primary visual cortex–essentially, the first cortical stop for incoming visual information. The traditional view used to be that, because of their position at the bottom of the information-processing hierarchy, neurons in V1 respond only to very simple configurations of light patterns (e.g., specific spatial orientations or frequencies). That view is increasingly being challenged by anatomical data indicating that V1 receives projections from a variety of other cortical areas and functional data showing that V1 neurons respond to a variety of seemingly high-level properties .

For example, Shuler & Bear recently demonstrated that neurons in rat visual cortex anticipate the onset of reward, showing a ‘learning’ profile much like that of dopamine neurons in much more anterior regions (i.e., closer to the front of the brain). In humans, neuroimaging studies reliably observe a top-down effect of spatial attention on visual cortex activation: when you expect a stimulus to appear in a particular location, activation in much of visual cortex (including V1) increases. What’s striking about this effect is that it’s endogenous (i.e., it’s produced by the brain and not by sensory information). The amplification occurs irrespective of whether a stimulus actually appears on-screen or not; what matters is whether people are visually attending to a spatial location. Moreover, the effect is spatially specific. V1 is highly retinotopic, meaning that it preserves the spatial structure of light hitting the retina quite well. By ‘stretching’ V1 out and visualizing it as a flat surface (normally it wraps around the banks of the occipital sulcus), researchers can actually pinpoint the location of stimuli in the environment based on the neural response. It turns out that the top-down modulation of V1 is strongest in the area of cortex that corresponds to the attended location in space.

The fact that attention is capable of modulating even the earliest reaches of visual cortex is pretty remarkable. Still, it’s at least intuitive that paying attention to a particular location would facilitate visual processing there (much of the early neuroimaging research on spatial attention was motivated by Posner and colleagues’ behavioral work demonstrating the existence of large facilitation effects). And it’s not exclusive to V1 by any means: the higher up in the visual hierarchy one goes, the stronger the facilitation effect gets.

The finding Jack et al. report in Neuron is just as intriguing, but much less intuitive. Essentially, they’ve discovered a signal in V1 that’s tied specifically to what they refer to as ‘task structure’. That is, the signal is produced whenever a task-relevant event occurs–e.g., a motor response. How did they identify the signal? Here’s what their first experiment looked like:

As you can see, there are two conditions in the task: immediate response and delayed response. In both cases, a checkerboard pattern is presented in 50% of trials and no stimulus s presented in the other 50%. However, in the immediate response condition, participants respond almost as soon as the stimulus period is over (their task is simply to judge whether the checkerboard was present or not), whereas in the delayed response condition there’s an 8 second delay before the response. That separation allowed the authors to compare the timecourse of visual cortex activation across the two conditions to detect any changes in signal. Here’s what they found:

The solid lines in this figure represent trials when the stimulus was actually presented; the dashed lines represent trials where there was no change in the visual display. Clearly, none of the effects depend on the actual presence of the stimuli, so I won’t say anything else about that. (Ignore the line colors for the moment; I’ll get to that in a second).

The figure pretty much tells the story. The left panel shows the V1 response when the subject’s motor response immediately follows the expected stimulus. As you can see, there’s only one peak, probably because the two events (stimulus and response) are so close together than fMRI isn’t able to tease them apart. Presumably, this lack of temporal resolution is also why (amazingly!) no one else had detected this effect before (people don’t t normally use 10 second trials just for the hell of it).

In contrast, if you look at the right panel, you can clearly see two distinct peaks of activation. The first is locked to the stimulus, the second to the motor response. The remarkable aspect of this finding is that there’s no change in visual display when the motor response occurs. The response period is cued by an auditory tone, so the second peak doesn’t reflect anything that’s happening on screen.

The fun doesn’t stop there. In addition to the temporal dissociation between the two V1 signals, Jack et al. also found two more ways in which the signals differed. One is apparent if you look at the colors in the right panel. The blue lines represent foveated stimuli (i.e., stimuli that are presented right in the center of the display, where vision is most acute) and the red lines represent stimuli presented in the periphery. What you can see is that the attention-induced V1 response (the first peak) is, not surprisingly, largest in the location where the subject expects the stimulus to appear. In contrast, the response-locked second peak is stronger in the more eccentric regions of V1 corresponding to peripheral spatial locations. In subsequent experiments, Jack et al. show that these effects are consistent: the stimulus-locked signal occurs wherever subjects expect the stimulus to appear, and the response-locked signal is always stronger in the periphery.

A second dissociation isn’t apparent in the figure above, because I removed the other panels (showing the timecourses in other areas of visual cortex) for simplicity’s sake. But the general finding is that, whereas the stimulus-locked signal is stronger in higher visual regions, the response-locked effect occurs almost exclusively in V1. That’s a pretty dramatic result in and of itself, since it suggests the latter signal is driven by bottom-up influences, and isn’t attentional.

The rest of the paper (including 5 other experiments) basically consists of a very meticulous attempt to rule out alternative sources of the response-locked V1 signal. I won’t go into detail, but here’s a short list of what Jack et al. show: (a) the signal isn’t just locked to overt motor responses, since it occurs even on ‘no-go’ trials when subjects aren’t supposed to respond; (b) the signal isn’t due to the auditory tone signaling the response phase (there are known projections from auditory cortex to V1, so this is a potential confound), because it occurs even when subjects count silently to themselves for 8 seconds before making self-generated responses; and (c) the signal isn’t due to blinks or eye movements, because it holds even after controlling for these factors. Suffice it to say that the effect appears to be highly reliable, doesn’t depend on obvious (and some not so obvious) confounds, and is independent of the previously identified top-down signal throughout visual cortex. In short, methodologically, the paper seems impeccable.

Of course, to many people, the question is: what does this signal mean? Here, Jack et al. don’t have very much to say. But then, it’s hard to blame them: as they observe themselves, what they’ve identified is, paradoxically, a signal in the earliest and most topographically organized part of visual cortex that’s nonperceptual and spatially diffuse. One possibility the authors reject is the notion that the signal might reflect a generic process such as arousal. They note that regions outside of visual cortex don’t show the same effect, and that the signal’s intensity doesn’t seem to track arousal-related variables such as task difficulty. Instead, Jack et al.’s major positive suggestion is that the novel V1 signal could potentially demarcate event boundaries, acting as a relatively general gating signal that helps the cognitive system transition between discrete states. It’s a pretty speculative idea, but the dynamics of the signal are so counter-intuitive that it’s hard to think of plausible alternatives. Given the impact that the study will probably have, odds are it won’t be long before a flurry of follow-up articles come along to hopefully provide more insight.

What I think is really neat about this study (beyond the general ‘that’s nifty!’ factor) is the number of different literatures it bears a relation to. Some of these are alluded to by the authors in the article, e.g., recent work on ‘mirror’ neurons. Mirror neurons are populations of neurons that increase their firing rate both when a specific action is performed and when the same action is passively observed in another person or animal. Mirroring systems have been already identified in humans and other primates for a range of behaviors (e.g., certain areas of premotor cortex are responsive to both performed and viewed hand actions), but the implication here is that even mice have at least a rudimentary mirroring system that can enable something like the social facilitation of pain. From an evolutionary standpoint, this makes perfect sense, of course. A mouse that feels a certain measure of distress when it sees a conspecific suddenly writhe in agony is a mouse that’s probably going to be a little more vigilant for threats in the environment, enhancing its odds of survival in potentially dangerous situations. But it’s a long way from evolutionarily plausible stories to hard data. What Langford et al.’s data suggest, along with other emerging evidence, is that some degree of mirroring may be a relatively primitive feature of the mammalian cognitive architecture.

The Langford study also breathes new life into the perennial philosophical question as to whether animals have feelings. Over at Pure Pedantry, Jake Young suggests that whether or not you think mice have empathy depends on your definition of empathy. Specifically, he focuses on the role of abstraction:

If your definition is that empathy is any commensurate change in behavior response to the perceived feelings of the other animal, then, yes, I would agree that the mice are feeling empathy. … On the other hand, if you define empathy as the ability to abstract another individual’s point of view from sensory data about them, then I don’t think this proves mice have empathy. It is clear that mice do not perform this level of abstraction. Mice could not, for instance, feel empathetic about another mouse from a story they heard about it.

I think this is an interesting way to frame the issue, but I think it also omits a third potential definition, which I suspect is the one most lay people hold–namely, that empathy is a specific kind of subjective experience that tends to make you want to act in certain ways. Naturally, this definition makes direct measurement of empathy impossible (which may be why Young doesn’t mention it): we don’t know what (if anything) mice feel, so we have to rely on their behavior as a rough index. That wouldn’t necessarily be a terrible thing, except that in practice, people (scientists included) often apply very different evidentiary standards to animals than to incapacitated humans or infants who may show a similarly restricted range of behavior. When a dog bares its teeth and growls, we refer to it as aggressive behavior, but many people are loathe to say the dog actually feels angry. In contrast, when a 6-month old infant cries and flails its arms, we don’t hesitate to make the latter attribution, even though the baby is just as incapable of expressing its frustration in a more abstract way.

This double standard matters for a couple of different reasons. For one thing, if we don’t apply consistent standards, we risk making the hypothesis that animals have feelings practically unfalsifiable. If the only evidence we’d accept for the existence of feelings in animals is verbal report or some other symbolic representation of emotion, we’re obviously out of luck. Now this isn’t necessarily a bad thing; but if we adhere to this position, we should at least have a reasonable explanation for why it’s ok to sacrifice a dozen monkeys in an electrophysiological study but not to euthanize a 2-month old infant. (Note: I’m not suggesting the two are equivalent, and I’m not arguing that animal testing is a bad thing. My point is just that one’s position on the subjective lives of animals has important implications for other domains, and these shouldn’t be brushed off just because they might be unpleasant to think through.)

A second, even more important reason we should seriously contemplate the notion that animals have feelings much like our own (though presumably less elaborate): failure to attribute emotions to animals is likely to result in the view that there’s something special and mysterious about human emotions, something that can’t be explained by studying animal models at all. And that would have a serious and negative impact on the scientific study of emotion.

Notice that most neuroscientists would find an analogous view absurd if applied to other domains of cognition. No one doubts that most other mammals really see the world roughly the way we do (subject to differences in sensitivity to various wavelength, etc.); no one questions the fact that studying the visual systems of cats and monkeys has taught us an incredible amount about the human visual system. And yet, the notion that emotions should suffer the same scientific fate somehow seems counterintuitive to many people. Neuroscientists who have no trouble discussing what their animals see, hear or sense in published articles get cold feet when it comes to emotions. I think no one’s summed up this sentiment better than Jaak Panksepp, an eminent neuroscientist now at Washington State University. Here’s a quote from a recent article:

The current neuro-behaviorist agenda, ascendant in the 1980s, led to a psychologically impoverished view of ‘‘emotions’’ in behavioral neuroscience, while yielding an enriched understanding of how emotional learning occurs in limited areas of the brain such as the amygdala (LeDoux, 1996). Most emotional processes that actually exist in animal brains have been disregarded (see Blanchard et al., 2001a for a sampling of modern behavioral neuroscience views of emotions). The failure of neuro-behaviorists to accept a diversity of emotions and the corresponding affective experience as a key aspect of animal brain functions has reduced the likelihood of useful cross-fertilization between animal and human studies.

Joseph LeDoux, the best funded animal emotional–memory researcher in America, publicly related how he failed to obtain approval for his initial grant applications until he extracted the term ‘‘emotion’’ from his proposed work to study classical-conditioning of fear and replaced it with learning and memory terms (see Panksepp, 2002). Other neuroscientists interested in emotions had comparable, but more sustained, funding problems throughout the last quarter century.

I’ve been a huge fan of Panksepp’s work for a while now. I find his arguments cogent, his theoretical models much more sophisticated than most, and his experimental work meticulous. And yet, despite being relatively well-known (largely as a result of his 1998 book Affective Neuroscience), Panksepp’s work–which spans several decades and a broad range of affect-related topics–hasn’t been nearly as influential as one might expect.

I can think of at least three reasons for this. First, most of Panksepp’s work has proceeded at a much lower level than people interested in emotions are generally comfortable with. Unlike domains such as vision, where it’s been clear for a long time that the phenomenal structure of vision is very different from the structure of the underlying mechanisms (there isn’t really a little man in your head flipping channels on a big screen), many people still don’t seem to believe that studying emotions at a systems neuroscience level is a fruitful approach.
Second, to some degree, behavioral neuroscientists have a vested interest in differentiating between human and animal emotions and denying the latter, because it’s easier to operate on a subject you don’t think feels much of anything. And third, the current dogma in affective neuroscience is one of simplification and dimensionality. There’s a strong focus at the moment on dimensional models, e.g., Richard Davidson’s famous hemispheric lateralization model (approach-related emotions are associated with greater left PFC activity, withdrawal-related emotions with right PFC activity), or Russell and Feldman-Barrett’s affect circumplex (which holds that most of the variance in emotions can be captured by just two orthogonal dimensions of valence and arousal). The main appeal of these theories is that they’re simple. You can explain their rudiments in a couple of minutes, and they’ve produced an inordinate amount of empirical research over the last three decades. Unfortunately, they’re also (at best) incomplete. No dimensional model I’ve seen does a passable job of explaining how one gets from 2 (or 3, or 4) dimensions to the full range of emotions one can experience and reliably distinguish between without a lot of vague hand-waving.

In most respects, Panksepp’s work is the antithesis of all of this. His models are extremely complex, drawing on an intricate knowledge of low-level animal circuitry; his avowed position is that human emotions depend largely (though certainly not entirely) on mechanisms present in other mammals too; and he emphasizes a ‘basic emotions’ perspective that argues that emotions are best understood in terms of a small number of core biological systems each associated with several emotions rather than abstract low-dimensional psychometric models. As a result of this contrarianism, his research has been relatively marginalized. (The fact that his writing is extremely opinionated and critical of other work probably doesn’t help, either.)

What does all this have to do with empathy? Well, personally I think Panksepp’s approach is the right one. And I think there are few enough examples of affective neuroscientists adopting the view that continuity exists between human and animal emotions that I think it’s noteworthy when a study is published that offers substantial support for that position. I think that studies like Langford et al. are fascinating in no small part because they provide a tantalizing glimpse into the neural substrates of emotion and offer us the hope that feelings aren’t intractably complex or unique to our species. They suggest the possibility that even a complicated emotion like empathy could potentially be explained by appeal to relatively simple properties of our cognitive architecture–properties potentially shared even by dogs, cats, and mice.

Based on Langford et al., one can tentatively suggest the possibility that appeal to a mirroring system might be all that’s necessary to explain a large subset of episodes we’d have no trouble labeling ‘empathy’ on a day-to-day basis. For example, take the immediate emotional reaction most people experience when they see someone sitting alone and crying. How different are such cases from mice writhing in pain? Well, a certain part of most people (myself included) immediately wants to shout “Very different!” But it’s not the scientific part; it’s a part that simply finds it hard to believe a relatively simple, seemingly reflexive behavior in a rodent could have much in common with the rich, complicated emotion we’ve all experienced first-hand. Ignore that visceral response and you can see the parallels. In both cases, what you really need is a mechanism that allows visual information to activate emotional pathways in much the same way they’d be activated by the corresponding first-hand experience. The rest falls into place given the prior existence of more basic pathways for various forms of pain processing. Far from being ludicrous, from a scientific standpoint, this hypothesis is an elegant and parsimonious one that could easily generate novel testable predictions. (For instance, electrophysiological recording could be used to attempt to identify the locus of the effect in mice, and the results then compared with human neuroimaging data in similar paradigms).

Now, I certainly don’t want to give the impression that I really think empathy in humans really is no more complex than in mice, or that the Langford paper offers a conclusive take on the matter. I don’t, and it doesn’t. My point is just that people–yes, even scientists!–have a deep-seated tendency to view emotion through different lenses than other domains of cognition, and it colors the way we approach our research. What in other domains would be seen as a clever experiment that opens the door to interesting avenues of research runs the risk of being dismissed out of hand in the field of emotion research as speculative and fantastic. And that would be a shame, since it is a clever experiment that opens the door to interesting avenues of research.