Hello Everyone,
Thanks so much for your thoughtful posts!
Please read the Menon & Kim (1999) article on cognitive neuroimaging and write a discussion board post by 9PM Monday, September 12th. This article was chosen by Gavin Bidleman as a primer to start to understand a neuorscientific approach, which he will discuss in more detail on Wednesday and tie into replication.
Please reflect on the approach with reference to replication, e.g., what novel insight might a neuroscientific approach provide when studying topics in cognitive science? This is another article in which I hope you focus on the big ideas without getting lost in the minutiae of it (although this may be a more palatable article compared to last week).
Please post your discussion as a response to this article. Afterwards, feel free to reply to others' posts. There are plenty of interesting ideas in this article; I am interested to discuss them with you at 3PM on Wednesday, September 14th.
Hope you are all well!
Best,
Dr. Braasch
ReplyDeleteTo begin, I simply wanted to point out that I really enjoyed the tone in which this article was written. Though I wasn’t a fan of the layout as I often found the flow of the paper a bit disjointed, I really enjoyed how the this paper taught you first and then dove into details later—a stark difference from Elman who assumed a certain level of expertise prior to engaging with his work. I understand that this was a review article and therefore that was the article’s job, however, that doesn’t mean that a job well done shouldn’t be commended. I found it especially helpful—especially for a paper that discusses neuroimaging—that the images presented were explained so that as a reader you were not left sitting there going “this looks pretty but I have no idea what this means”.
Moving on to our broader topic of replication, I feel replication in regards to neuroimaging is a fascinating discussion in that neuroimaging is widely used and the findings are widely genaralizable, yet the degree of individual variation between neuroimages must be VAST. The article mentions that physiological variations are very difficult to control for and may create noise within the images. That raises the question, can any neuroimaging technique truly be replicated then? I imagine with the nature of this research that we are discussing direct replication and as we have mentioned before, as soon as you change the subject pool it already becomes more difficult to obtain a direct replication and that is when a researcher is attempting to replicate a psychological finding, never mind a physiological one.
Controlling constants such as physiological fluctuations must be paramount to replicating neurological findings and I truly applaud neuroscientists for literally controlling for human variation—and we cognitive scientists think it’s hard to control for prior knowledge, never mind neural-hemodynamic coupling constants.
As I was reading the article, I thought of a few questions that might stimulate discussion posts. When you post your response, feel free to think about and respond to any of these issues (and of course any other issues you think are useful to bring up).
ReplyDelete1) What affordances does a neuroscientific approach have for studying cognition? Said another way, what nuances of cognition can neuroscientific approaches capture that can't be captured otherwise given more basic behavioral studies?
2) What limitations does a neuroscientific approach have for studying cognition?
3) Are there any drawbacks in using a neuroscientific approach to study cognition?
4) How might neuroscientific approaches to studying cognition be tied to replication?
5) Are there any examples, from your research or research you know about, in which neuroscientific approaches to studying cognition may be used as a method for replicating prior work? What predictions would you make?
In comparison to the Elman article, this week’s reading was a nice contrast in terms of layout and content. Although I was slightly thrown off by the diagrams and images at first, I felt a much stronger understanding of this week’s article. I did feel like there was a similarity between the articles when discussing the chronometry of the brain in Box 1. Overall, this part article was getting at the issue of viewing connectivity in a neural network in a parallel, feedforward and feedback loop system, which reminds me Elman’s issue with viewing language processing in a similar way. It appears that these elements of cognition and neural functioning are best viewed in a much more dynamical approach.
ReplyDeleteI really liked the questions that Dr. Braasch posted and felt they aligned with some of the questions I was thinking of when reading. In terms of using this approach to study cognition, I feel there are several benefits to studying the physiological elements as well as the behavioral elements of cognitive processing. While I cannot explain it as well as the article, it appears that by combining both physiological and behavioral measures allow for a mapping of the desired functions. This mapping allows for the insight into what neural activity is associated with what processing strategies. Overall this concept is so intriguing to me, but I cannot imagine the difficulties that neuroscientists face when trying to achieve these results or in terms of replication.
I have a personal research interest in critical thinking and knowledge sourcing, and immediately began to wonder how techniques such as fMRI could play a role in studying these concepts. Naturally, one could wonder if there are external stimuli causing in internal response we cannot account for, during a critical thinking tasks it would be interesting to see what types of environments, inventories, and experimental designs impact the physiological aspects of critical thinking. As far as knowledge sourcing, I’m a little stumped as to how monitoring neural activity could specifically aid in understanding what sources they chose to accept or why they accept certain sources. Perhaps there are certain internal responses that in combination with behavioral responses could allow for a deeper understanding of how or why behind a person’s knowledge sourcing? I know the answer is there and probably very obvious, but for now I’ll leave that for the neuroscientists.
The technical aspects of this article were a bit beyond me, but I did find many of the main concepts, relationships, presuppositions, and findings to be interesting. I'm going to save some of my concerns for discussion, since Lauren Dahlke and I are leading this week, but I have plenty and no shortage of verbosity!
ReplyDeleteThe last article took a dynamical approach to language learning. I mentioned in class that Brooks does the same for motor tasks (robotics). Many philosophers are moving towards understanding cognition dynamically, often in terms of differential equations. The author's of this week's article are also taking a dynamic approach. I wonder if these examples illustrate a movement in cognitive science towards dynamical explanation, or whether these just happen to be outliers (or perhaps the views of persons outside the field which are not representative of that field). I also am curious about the relationship in this article, in particular, between "single event" research and dynamic approaches, as well as "static" approaches. Furthermore, I wonder if non single event studies count as a kind of replication. It seems the "subject" changes, especially if we account for time and memory, but it also seems that many aspects of the experiment are kept constant.
There was a quick reference to primary motor activation during the "preparation" phase of certain motor tasks. This is interesting, especially in the context of mirror neurons. It seems that whether I imagine doing something, do the thing, imagine someone else doing the thing, or see someone else doing the thing, many of the same areas of the brain activate. Non-overlapping areas seem to correspond to whether the action was perceived or imagined, and whether it was mine or someone else's. This is true for motor action as well as expressive behavior generally (e.g., smiling as displaying happiness).
There is a supposition that the brain is functionally segmented, or that certain areas do certain things. Recent work in neuroplasticity and a few case studies may challenge this supposition, though. If this supposition falls, what falls with it?
Overall, I believe Menon and Kim did justice to explaining some of the spatial and temporal limits of using fMRI, and delving into possible ways to overcome some of these limits. What struck me most, however, is that it seemed as though they were trying to force a way for fMRI to be a better temporal measure of neural activity, when it is by its nature an indirect measure of neural activity and has poor temporal resolution. I do agree that their single-event paradigm may be a way to increase temporal resolution, but the authors weren’t quite as eloquent as Elman in regards to indicating that this paradigm is by no means a solution to the temporal problem fMRI presents.
ReplyDeleteThis brings me to a proposal regarding replication using the technique described to perhaps get a better idea of the relationship between neural activation and the hemodynamic response measured by fMRI. As one of the focuses of the paper is the temporal limitation of fMRI, I found myself asking, “What if they could combine measures for the best of both worlds?” By this, I mean, why not combine electroencephalography (EEG; which has great temporal resolution, but poor spatial resolution) with fMRI (which has the opposite issue)? There are MRI-friendly EEG systems, and I think (perhaps naively) they could be used in conjunction to gain more information about both when activation starts and where it is, and how the electrical signals of the brain are related to the hemodynamic response. I’m sure this method has been used, so perhaps in combination with single-event trials described by Menon and Kim, we may be able to get a clearer view of brain processes, with a higher signal-to-noise ratio. Of course, this proposal in itself has its own limits. Still, it is something that could be explored.
I found this article very helpful, as I had little to no previous knowledge of how fMRIs work. I knew they were used to monitor brain activity, but wasn’t really sure how they did that. It was good to read the easy to understand explanations in this article. Though I didn’t much care for the layout of the article, what with the figures cutting pages in half, I did enjoy most of it. Those same figures that I found irritating turned out to be really interesting to go back and read once I had finished a section.
ReplyDeleteI found both the capacities as well as the limitations of fMRI to be fascinating. Much of the information in this article still went over my head, but I found the tables to be very helpful in understanding. It’s amazing that we can see in almost real time what is going on in tiny areas of the brain. It really opens up our ability to study exactly what is going on neurologically when a subject is performing a task. Since I’m very interested in behavioral neuroscience as well as cognitive neuroscience, this is great news for me! I’ve only ever used EEGs in the past, but maybe hopefully someday in the future I’ll get to run studies using fMRIs. One thing I found especially interesting was that there may be biological limit to how exact a picture of the brain we cn achieve, due to the nature of blood flow in the brain. This limitation may mean that we cannot see exactly what is causing what in the brain, since all the interplay between different areas may be large, electrically, but immeasurably small in terms of blood flow.
This article provides a good introduction of how a neuroscientific approach works. Even though the neuroscientific terms in the article are overwhelming to me, it finally makes sense to me after I looked up every term and reread the article. From my knowledge background of cognition, I learned and understand cognition in the general learning process, including some basic assumptions of information processing theory in learning. These basic information processing model that I learned from other course helps me form a structure of how people process information, Menon and Kim’s article helps me get to understand what is going on within one part of the structure, which is very helpful because it is every necessary to know how cognitive task is perceived, processed and responded if I want to explore what influence learning.
ReplyDeleteAs regard to the limitations in using fMRI, I think the neuroscientific approach may be easy to replicate in the lab setting, but I wonder whether the results may be the same in the real life situation (may because my lack of understanding on the topic). Because in the end of the article, the authors mentioned the technical impediments such as the noise sources intrinsic to the measurement process. I mean, even in the lab, there are some technical impediments when measuring MRI, how people process certain kinds of information in different environment or situation may be influenced by many sources. So I feel it is very hard to measure the influence from environment or situation when exploring the cognitive processing.
The article this week discussed the spatial and temporal limits associated with using fMRI to conduct cognitive neuroimaging. In overview, portions of the brain “light up,” or have increased blood flow, when they are engaged. fMRIs can map the flow of blood during single events, meaning a single cognitive, sensory, or motor event like a single finger movement. One problem, however, is that the images produced by the fMRI is a static representation of a dynamic process. On page 210 it reads “Despite the ability to make MRI images on a modern MRI scanner in excess of 20 per second, the temporal dimension of fMRI has not been exploited…” Were scientists hesitant to break down measurements because they would still just be averages of the mental processes occurring? I don’t understand why they wouldn’t want to explore that facet more deeply. The article was published in 1999, though, so I hope they have changed their tune. A limitation of the present article is that fMRIs are able to track blood flow, but that doesn’t give us much information on how the neural processes themselves work, and the order of activation.
ReplyDeleteA point briefly mentioned on page 209 is that while some neural substrates are activated in a certain order, there is evidence that others are not. The brain can process many stimuli in parallel, which might be a limitation of the fMRI (but I’m not a cognitive person so I’m really not sure!).
The repeated trials aspect of fMRI studies is in a way a direct replication of the original research question. I would think it would be fairly easy to come up with conceptual replications, as all you would need to do would be to change the stimuli presented. As fMRI technology changes, too, more and more information would be revealed from each trial.
One thing that interested me is that (I’m guessing) all of the studies mentioned were conducted using adults, or at least children old enough to be able to stay still in the fMRI. That calls into question how studies like this could be conducted on infants and small children, who are still forming their neural networks, and might have different results.
This comment has been removed by the author.
ReplyDeleteMenon’s article gives a very good introduction of cognitive neuroimaging with functional magnetic resonance imaging (fMRI). These days’ fMRI gives a new dimension in cognitive and basic nonscientific research. Now because the idea of fMRI researcher can analyze human perception and cognition more indepthly. On the top of that I feel like fMRI is more advance technology based research so interms of global perspective still many countries are still behind in terms of using this high tech neuroscientific research.
ReplyDeleteWhat we are measuring using fMRI? A quick response of this question is, it measures the increase in regional cerebral blood flow(rCBF). At the same time fMRI does not directly detect the electric activity nor does it measure the rapid increase in metabolism.
I have found a interesting relation between cognition and neuroscientific research in this article. In cognitive science we look for information processing and same thing fMERI do with a combination of spatial-temporal scale. One limitation that I have found in this study is there might be a significance difference in the results of this study due to blood circulation change in the brain. Because MRI took the image of the brain based on flow of the blood in the brain. Overall I enjoyed reading the article and it is worth reading for since my interest is epistemic cognition, where I see the information processing in human brain more visually.
This article was a breath of fresh air after Elman’s article last week. Although I am no expert in neuroscience, I felt this article did a much better job explaining the terminology and “dumbing it down,” if you will.
ReplyDeleteI see many limitations to using a neuroscientific approach to studying cognition. Although these methods are extremely advanced, I believe that they have the ability to overlook some of the simpler aspects of cognition, or those that we don’t need a fancy machine to examine. I do, however, think that this technology is incredible and will take neuroscience to new levels. I think it is important for any scientist to take caution when using new technology and to remember the basics.
Similar to what Jeff said, I think the individual variance in brain imaging techniques must be incredibly wide. It seems as though neuroscience is making huge strides in being able to map out certain cognitive processes in the brain, but how do individuals differ? Is there any way to truly generalize with such a high level of variance? With a basic experiment, we can for the most part assume that participants are behaving in the same manner. But what about what is happening in their brain? That is the type of research I would like to see done- research focusing on replications across broad and different samples of people. I think focusing on these individual differences could lead to new and exciting findings in the realm of neuroscience.
Having taken neurobiology in undergrad, working in an emergency room at a hospital for 2 years, and taking numerous other biology classes, I had a bit more of a background understanding of some of the more technical aspects described in the text. With that being said, I thought that the application of fMRI and other neuroimaging techniques in understanding chronometry was an extremely interesting and novel idea. It provides details that basic behavioral studies could not hope to discover—as behavioral studies are not formatted to give you information about the complex activation of the brain by looking at microvasculature. The timing of the neural activation, location, and the level of intensity of the blood flow all can delineate important distinctions between tasks of varying difficulty. This is precisely why neuroimaging serves as a perfect example of how conceptual replications can confirm and give further detail in regards to understanding theory.
ReplyDeleteThe neuroscience approach might be limited for studying cognition in that it doesn’t touch on the “why” of some of these aforementioned behavioral studies, and biologically reducing cognition seems to lose part of the qualities that we normally think that cognition has. I’m curious about the sorts of progress that has been made since this article has been published, seeing how it was published nearly 20 years ago by this point.
I would be specifically interested in seeing how fMRI has been used to analyze people with mental illness. Such things as looking at stereotypies, language incoherency, and neurological processes that occur during psychotic episodes might be very elucidating. There are numerous applications that can still be applied (and maybe already have been), but current technologies are still not at a level that clearly illustrates cognition from a global vantage point.
I found the article this week to not only be easier to read than last week, but also very informative considering my limited background with neuroscience. It seems as though one huge affordance a neuroscientific approach to cognition is a physiological response to stimuli. Traditionally when studying cognition, competent researches rely heavily on participant behavior. With a neuro/traditional approach, researchers are able to look at both behavioral response as well as physiological response which can be useful as a mapped second layer of cognition. I found myself coming the conclusion that these approaches, when correctly handled, could be conceptual replications, but this might be a far-fetched idea.
ReplyDeleteA limitation I foresee is trouble in attributing specific neurological responses with specific stimuli. I see this particular problem arising from variance between images taken within and between subjects. Replicating an image at a precise moment in some determined cognitive process in one individual seems near impossible to me. I also think that variations in the exact structure and physiological response between individuals would be nearly impossible to determine. However, I am highly unfamiliar with this topic so perhaps it can be or has been done already. It seems to me that controlling for these levels of variation would be incredibly difficult. Therefore, I would assume general trends would be more likely to be found but precise replication would be difficult.
Reading through the article, it seems that one of the primary strengths of a neuroscientific approach is it provides physiological indicators of psychological phenomena, which is useful for a couple of reasons. First, examining these indicators provides some additional level of observability to inherently abstract processes. While there is good support for a number of theoretical claims about cognition, one of the limitations is that cognition cannot be observed directly, resulting in reliance on indirect, often behavioral, indicators to measure cognitive variables. Having additional, albeit limited, evidence of aspects related to cognition provides incredibly helpful complementary information. Second, such data can be used to make theories in cognitive science more robust. Because neuroscientific approaches help get at potentially more generalizable processes and underlying structures, the information garnered may help researchers connect phenomena in ways that may be impossible using behavioral data alone. One may notice similarities in neurophysiological patterns during tasks that may go overlooked when using only behavioral reactions. Further, with adequate ability to consider neural activation underlying cognition over time, one may be able to see differences associated with cognitive processes that would not be discernible in higher-order reactions. In general, neuroscientific approaches provide an additional means of expanding theory, making more accurate hypotheses, and providing alternative evidence for assumptions in cognitive science, therefore serving as a type of conceptual replication.
ReplyDeleteWhile neuroscientific approaches may provide the benefits previously mentioned, they retain a number of limitations. As mentioned in the Menon and Kim article, there may be natural barriers that limit methods of measuring neurophysiological variables, as is seen with fMRI. In this case, there may be hard limits on the information one might glean from certain techniques. The techniques and equipment used also pose some problems for replication, particularly when variables may or may not be examined as thoroughly depending on the strength and quality of the technology used. This is illustrated by the comments in Menon and Kim stating that certain results should be producible with more accessible technology and methods, though this depends on a number of factors that one may or may not have known to consider. Results of neuroscientific studies may also be obscured by individual differences, thus the reason the article alludes to the use of aggregated data in such research. The degree to which these differences adversely impact findings would naturally rely on the complexity of the processes examined. One would likely be able to make broad claims with confidence (e.g. which regions are generally associated with which cognitive processes, which sequences are generally followed, etc.), while making specific claims might be exponentially more difficult. In this regard, the issue mirrors a similar one in physics and other fields, in that predictions using larger-scale variables are far easier to make than those for variables of small scale (i.e. Newtonian vs. Quantum levels of physics). This naturally has implications for the predicted success of replication. Finally, one other potential issue with neuroscientific approaches is that they may be interpreted in reductionist ways. In these cases, one runs the risk of assuming that understanding certain strings of variables equates to understanding the related integrative cognitive processes of interest. However, such a line of thinking may undermine more dynamic understandings of cognition and oversimplify incredibly complex interactions.
Although, by nature of discussing this, I am conceding that I viewed the comments prior to forming my own responses, Jason has raised some questions that are personally extraordinarily relevant and I have been trying to struggle through since I began my undergraduate honors thesis work. The first one is this:
ReplyDelete“Are there any examples, from your research or research you know about, in which neuroscientific approaches to studying cognition may be used as a method for replicating prior work? What predictions would you make?”
For my UG thesis, I wanted to combine my advisors work, the misinformation effect, with research on dual-process theories of reasoning. For a quick overview, see Evans (2003) (this idea was also popularly portrayed in Daniel Kahneman’s Thinking Fast and Slow). For a more thorough, yet very verbose explanation, see De Neys (2013) or Pennycook et al. (2015). Essentially, this theory posits that we have two ways of processing information, a quick style, based on heuristic and associative cues (System 1) and a slower, more deliberative, yet cognitively demanding, system (System 2). Underlying the various arguments surrounding this theory is the question of how the various systems operate: in parallel? Through conflict detection? Through “metacognitive cues” (a good idea, but not useful in determining actual biological structures of these systems)?
At present, the conflict monitoring hypothesis seems to be the most well-supported and has led to the development of a model, by Pennycook et al. (2015), which explains some of the ambiguities which the literature has neglected so far. The largest of these is the question of what exactly a conflict is (neurologically) and how it triggers System 2 processing. A key recent development was Pennycook et al’s proposal that triggering System 2 does not necessarily mean a “correct” response. That is, individuals can engage in deeper processing of information and still maintain (or even strengthen) their biased response. This is especially true when dual-process theory is applied to beliefs (e.g. politics, religion), as it commonly is. While this is somewhat supported by Pennycook et al’s studies, this is a case where neuroscientific evidence could almost end this debate regarding how these two systems operate.
To test this, there is a fairly straightforward design, close to some work which has already been conducted. One could take a pre-test measure of individuals’ beliefs on a given topic (e.g. the age of the Earth) and present belief consistent or inconsistent statements/texts (e.g. prompting young Earth creationists with a short statement regarding geologic time) and ask them to agree or disagree with the statement. Furthermore, System 2 could be activated without the subject’s awareness, fairly consistently, by using one of the manipulations commonly used to elicit deeper processing (e.g. the use of disfluent text).
The neurological correlates of conflict detection and System 2 activation are fairly well-established (generally, it involves heightened activity in the anterior cingulate cortex and right lateral prefrontal cortex), so it would be fairly easy to determine that System 2 was engaged in these individuals, while monitoring whether they gave a belief-consistent vs. inconsistent response. If it could be shown that system 2 processing is engaged, even in the production of a belief-consistent response, this would lend overwhelming support to this model.
In the case of dual-process theory, there is a great need for the interjection of some neuroscientific evidence to support the various mechanisms which are proposed to operate. There are many very logical and empirically-validated explanations, but this is ultimately a question of which explanation maps most neatly onto the physical structure of the brain.
The links I hoped would hyperlink in the text:
DeleteEvans (2003)
http://faculty.weber.edu/eamsel/Classes/Methods (3610)/Old Sections/Fall 2010/Fall 2010 Project/Evans (2003).pdf
DeNeys (2013)
http://www.wdeneys.org/data/TAR reprint special issue.pdf
Pennycook et al. (2015)
https://www.researchgate.net/publication/278404209_What_makes_us_think_A_three-stage_dual-process_model_of_analytic_engagement
As the technical aspects of this are a little too advanced for me to really assess off just one article, I'd like to mainly focus on a couple lingering questions I had once I finished this article. It was mentioned that brain imaging relied upon an understanding of the brain as highly structural and specialized. While I'm sure there may be plenty of information supporting such a theory, I wonder if fMRI studies merely reinforce that view without considering alternative theories.
ReplyDeleteAlso, I have to imagine there are serious challenges to replicating fMRI work. It may be an amazing tool for a physiological investigation of cognition, but it certainly can't be a cheap method of analysis. While there are certainly groups and institutions in academia that have access to these tools to attempt replication, a high barrier to entry must mean there is less scientific expertise to go around in such an endeavor. From an outside observer, fMRI studies seem primed to 'pass on' experimental and interpretative errors due to their highly specialized nature.
I think that the idea of the replication process in neuroimaging research is interesting. Throughout the article, I noticed a few examples of studies done using different participants or methods/techniques to get at the same underlying research questions, which I saw as somewhat of a conceptual replication. For example, previous research that had been done on primates helped inform research on humans, and new neuroimaging techniques allow for replications with higher spatial resolution. With the advent of new technologies, it seems like replication in this field is extremely important, especially since conducting the same study with better technology can possibly lead to clearer, more specific results. This paper was written in 1999, however, so I would be curious to know how neuroimaging has continued to progress as a field up until this point in time.
ReplyDeleteStudying cognitive psychology, I can see a lot of potential applications where neuroimaging can be used in research. Not only is there potential for novel neuroimaging studies within psychology, but there is also the potential that already existing research and findings can be further examined using neuroimaging. Using neuroimaging to replicate and extend previous research could help to discover mechanisms behind psychological phenomena.
This paper describes how fMRIs have been used recently in experiments. This seems to be a good summary to prepare someone for reading a more technical paper with an experiment using an fMRI. (Why did they have to format this paper with these big boxes that break the flow?? It made it annoying to read.)
ReplyDeleteThe first thing I wondered is, how much noise and variation is there between trials? It is believable that the measured spikes are caused by the stimulus but I'm surprised that it is so apparent, considering how many other things are going on in the brain. There are many potential unintended stimuli (e.g., environmental sounds or an itch, or even a lack of stimuli) going on at any time, plus any thoughts the participant is having at any given time. The authors very briefly bring this up on page 5 but it seems like a really big deal. Another thing I'd like to see is just a longer term baseline to see how much the signal varies.
Another question I always have when reading work like this is whether there is any real understanding for what it means when a particular location "lights up". Researchers have been studying localization for a long time now (e.g., Jerry Fodor...) and try to make claims that if two separate activities make the same location light up, then it must be using the same processing unit... but that seems like quite a leap.
The reason I bring up both of these points is because they could have a huge impact on replication. If the localization of activity varies between participants or if unintended stimuli vary between trials/participants/replications then it could really change the results. It is my impression that all of this would be noise and the real signal would still be seen, but it is worth looking into!
Somewhat unrelated, this type of stuff is interesting to me because software engineering researchers have recently been doing experiments where they ask participants to do programming tasks while in an fMRI machine. The work so far as been very shallow without much to take away from it but I think it is a great step in understanding how programmers comprehend code (a whole subfield that psychologists and computer scientists have been studying since the 70s).
ReplyDeleteThis article gives good introduction about fMRI neuroimaging technique in terms of those from principle of measurement, spatial-temporal limitations to technical achievement so far. In this article, the authors usually focused on the limitation that conventional scanners quite slow to catch dynamic cognitive processing. But, as they addressed in this review, these temporal limitation seems to become solved as stronger measurement unit(T) MRI is developed. Comes to spatial limitation, it seems hard to overcome since it happens because of biological feature of human body. Alternatively, we can exclude these noise signal by manually after experiment.
Despite of couple of limitations, I think fMRI is still effective method for measuring brain activity corresponding to cognitive processing. It’s non-invasive, well-rooted on biological principles, directly showing how brain is working. Actually method that I usually worked on was EEG which is effective on millisecond-range temporal resolution, even better than MRI. I think, if possible, I could adopt these two techniques on my research interests such as correlation between language and knowledge, bilingualism, and semantic memory. With EEG, I can measure electrical signal of brain during processing written language and symbolic information (non-literal knowledge) so that comparing differences of each brain signals. Also fMRI would allow to see whether there are different brain regions involved processing first language and second language or not. Or, if I combine two techniques to these researches, I may obtain more solid results. Anyhow, I think these kind of techniques are also reliable on cognitive science field as much as behavioral approaches.