INTRODUCING COGNITIVE NEUROSCIENCE 9 41 Functional MRI Naturally MEG ERP PET Brain Map Column Layer Neuron Dendrite Single-cel recording Synapse -1 0 1 3 4 67 Second Minute Hour Day Temporal resolution (log seconds) FIGURE 1.5:The methods of cognitive neuroscience can be categorized according to their spatial and temporal resolution. Adapted from Churchland and Seinowski,1988 physiological changes associated with blood supply to the brain,which evolve more slowly over time.These are called her odynamic methods and are considered in Chapter 4. The methods of cognitive neuroscience can be placed on a number of dimensions(see Figure 1.5): The temporal resolution refers to the accuracy with which one can KEY TERM measure when an event is occurring.The effect s of brain damage are Temporal resolution permanent and so this has no temporal resolution as such.Methods The accuracy with whicr such as EEG,MEG,TMS and single-cell recording have millisecond resolution.fMRI has a temporal resolution of several seconds that an event (e.g..a ph reflects the slower hemodynamic response logical change)occurs. THE DIFFERENT METHODS USED IN COGNITIVE NEUROSCIENCE Method Method type Brain property used EEG/ERP Recording Noninvasive Electrica Recording Invasive Electrical multi-unit)recordings TMS Stimulation Noninvasive Electromagnetic Stimulation Noninvasive Electrical MEG Recording Noninvasive Magnetic PET Recording Invasive Hemodynamic fMRI Recording Noninvasive Hemodynamic fNIRS Recording Noninvasive Hemodynamic
Introducing cognitive neuroscience 9 physiological changes associated with blood supply to the brain, which evolve more slowly over time. These are called hemodynamic methods and are considered in Chapter 4. The methods of cognitive neuroscience can be placed on a number of dimensions (see Figure 1.5): • The temporal resolution refers to the accuracy with which one can measure when an event is occurring. The effects of brain damage are permanent and so this has no temporal resolution as such. Methods such as EEG, MEG, TMS and single-cell recording have millisecond resolution. fMRI has a temporal resolution of several seconds that reflects the slower hemodynamic response. FIGURE 1.5: The methods of cognitive neuroscience can be categorized according to their spatial and temporal resolution. Adapted from Churchland and Sejnowski, 1988. 4 3 2 1 0 –1 –2 –3 –3 –2 5 6 7 –4 Brain Naturally occurring lesions Map Column Layer Neuron Dendrite Synapse Millisecond Spatial resolution (log mm) Day –1 0 1 2 3 4 MEG & ERP Functional MRI PET TMS Multi-unit recording Single-cell recording Second Minute Temporal resolution (log seconds) Hour THE DIFFERENT METHODS USED IN COGNITIVE NEUROSCIENCE Method Method type Invasiveness Brain property used EEG/ERP Recording Noninvasive Electrical Single-cell (and multi-unit) recordings Recording Invasive Electrical TMS Stimulation Noninvasive Electromagnetic tES Stimulation Noninvasive Electrical MEG Recording Noninvasive Magnetic PET Recording Invasive Hemodynamic fMRI Recording Noninvasive Hemodynamic fNIRS Recording Noninvasive Hemodynamic KEY TERM Temporal resolution The accuracy with which one can measure when an event (e.g., a physiological change) occurs
10 THE STUDENT'S GUIDE TO COGNITIVE NEUROSCIENCE KEY TERM The spatial resolution refers to the accuracy with which one can measure where an event is occurring.Lesion and functional imaging methods have comparable resolution at the millimeter level.whereas single-cell which one can measure recordings have spatial resolution at the level of the neuron. lipment is located where an event (e.g.,a Theofamethod reers to whether the physiological change)is internally or externally.PET is invasive because it requ s an iniection of med on the brain nan animals DOES COGNITIVE PSYCHOLOGY NEED THE BRAIN? As already noted,cognitive psychology developed substantially from the 1950s,using information-processing models that do not make direct reference to the brain.If this way of doing things remains successful,then why change? Of course,there is no reason why it should change.The claim is not that cognitive neuroscience is replacing cognitive psychology (although some might endorse this view),but merely that cognitive psychological theories can inform theories and experiments in the neurosciences and vice versa However,others have argued that this is not possible by virtue of the fact that information-processing models do not make claims about the brain (Coltheart,2004b:Harley,2004). Coltheart(2004b)poses the question: Has cognitive neuroscience,or if not might it ever (in principle,or even in ognitive FIGURE16:One could take practice).successfully used data from many different measures in a forced-choice response adjudicate bet ompeting inform on-processing models of som task:behavloral(reaction cognitive system)? nts)orb (p.21) (electromyographic [EMG] BOLD response in fMRI or EEG that one there are a be used to inform cognitive of thi can meas tha me ure in rd forced ice reactc theory. reac error rate sweating (skin (E) hemodynamic changes will relate to the task in some way and can be used to inform theories about the task. Eye-movements EEG To illustrate this point,consider an example.One could ask a simple question R such as:Does visual recognition of words and letters involve computing a representation R that is independent of case?For example, does the reading system treat“E'and“e”as equivalent at an early stage in processing or are“E”and“e”treated as different letters until EMG some later stage (e.g.,saying them aloud)?A way of investigating this using a reaction-time
10 THE STUDENT’S GUIDE TO COGNITIVE NEUROSCIENCE • The spatial resolution refers to the accuracy with which one can measure where an event is occurring. Lesion and functional imaging methods have comparable resolution at the millimeter level, whereas single-cell recordings have spatial resolution at the level of the neuron. • The invasiveness of a method refers to whether the equipment is located internally or externally. PET is invasive because it requires an injection of a radio-labeled isotope. Single-cell recordings are performed on the brain itself and are normally only carried out in non-human animals. DOES COGNITIVE PSYCHOLOGY NEED THE BRAIN? As already noted, cognitive psychology developed substantially from the 1950s, using information-processing models that do not make direct reference to the brain. If this way of doing things remains successful, then why change? Of course, there is no reason why it should change. The claim is not that cognitive neuroscience is replacing cognitive psychology (although some might endorse this view), but merely that cognitive psychological theories can inform theories and experiments in the neurosciences and vice versa. However, others have argued that this is not possible by virtue of the fact that information-processing models do not make claims about the brain (Coltheart, 2004b; Harley, 2004). Coltheart (2004b) poses the question: Has cognitive neuroscience, or if not might it ever (in principle, or even in practice), successfully used data from cognitive neuroimaging to make theoretical decisions entirely at the cognitive level (e.g. to adjudicate between competing information-processing models of some cognitive system)? (p. 21) Henson (2005) argues that it can in principle and that it does in practice. He argues that data from functional imaging (blood flow, blood oxygen) comprise just another dependent variable that one can measure. For example, there are a number of things that one could measure in a standard forced-choice reactiontime task: reaction time, error rates, sweating (skin conductance response), muscle contraction (electromyograph), scalp electrical recordings (EEG) or hemodynamic changes in the brain (fMRI)—see Figure 1.6. Each measure will relate to the task in some way and can be used to inform theories about the task. To illustrate this point, consider an example. One could ask a simple question such as: Does visual recognition of words and letters involve computing a representation that is independent of case? For example, does the reading system treat “E” and “e” as equivalent at an early stage in processing or are “E” and “e” treated as different letters until some later stage (e.g., saying them aloud)? A way of investigating this using a reaction-time FIGURE 1.6: One could take many different measures in a forced-choice response task: behavioral (reaction time [RT], errors, eyemovements) or biological (electromyographic [EMG], BOLD response in fMRI or electrical activity in EEG). All measures could potentially be used to inform cognitive theory. Adapted from Henson, 2005. By kind permission of the Experimental Psychology Society. fMRI Eye-movements EEG EMG RT KEY TERM Spatial resolution The accuracy with which one can measure where an event (e.g., a physiological change) is occurring
INTRODUCING COGNITIVE NEUROSCIENCE 11 RADIO ADIO MOUSE -RADIOL One genera nng in reacton-tme studl isthat tis faster to poces a stimulus if the same stimulus has recently been presented. For example if asked to make a speeded decision about RADIO (e.g.,is it animate or inanimate?),performance will be faster if it has been previously encountered Dehaene etal.(2001)investigated this mechanism by comparing reaction-time measures with functional imaging(fMRI)measures.In this task,the first word in each pair was presented very briefly and was followed by visual noise.This prevented the participants from consciously perceiving it and,hence,one can be sure that they are not saying the word.The second word was consciously seen and requires a response.Dehaene et al.found that reaction times were FIGURE 1.7:Both reactior faster to the second word when it follows the same word,irrespective of case times and fMRl activation Importantly,there was a region in the left fusiform cortex that shows the in the left fusiform region same effect (although in terms of"activation"rather than response time). der This is shown in Figure 1.7.In this concrete example,it is meaningless to argue that one type of measure is"better"for informing cognitive theory (to return to Coltheart's question)given that both are measuring different aspects of the same event.One could explore the nature of this effect further by,for instance,presenting the same word in different languages(in bilingual ed from Dehaene. Same word Different word fMRI activity 500 ms radio Same Time 29m80 220 29 ms RADIO 625 ■5 ame word 29m 620 271m☒ 615 610
Introducing cognitive neuroscience 11 measure is to present the same word twice in the same case (e.g., RADIORADIO) or different case (e.g., radio-RADIO) and compare this with situations in which the word differs (e.g., mouse-RADIO, MOUSE-RADIO). One general finding in reaction-time studies is that it is faster to process a stimulus if the same stimulus has recently been presented. For example, if asked to make a speeded decision about RADIO (e.g., is it animate or inanimate?), performance will be faster if it has been previously encountered. Dehaene et al. (2001) investigated this mechanism by comparing reaction-time measures with functional imaging (fMRI) measures. In this task, the first word in each pair was presented very briefly and was followed by visual noise. This prevented the participants from consciously perceiving it and, hence, one can be sure that they are not saying the word. The second word was consciously seen and requires a response. Dehaene et al. found that reaction times were faster to the second word when it follows the same word, irrespective of case. Importantly, there was a region in the left fusiform cortex that shows the same effect (although in terms of “activation” rather than response time). This is shown in Figure 1.7. In this concrete example, it is meaningless to argue that one type of measure is “better” for informing cognitive theory (to return to Coltheart’s question) given that both are measuring different aspects of the same event. One could explore the nature of this effect further by, for instance, presenting the same word in different languages (in bilingual FIGURE 1.7: Both reaction times and fMRI activation in the left fusiform region demonstrate more efficient processing of words if they are preceded by subliminal presentation of the same word, irrespective of case. Adapted from Dehaene et al., 2001
12 THE STUDENTS GUIDE TO COGNITIVE NEUROSCIENCE speakers).presenting the words in different locations on the screen and so on This would provide further insights into the nature of this mechanism (e.g. what aspects of vision does it entail?Does it depend on word meaning?) However.both reaction-time measures and brain-based measures could be ot the case that functional imaging is merety nction that has e cognitive som extent this is true.But the itten computer ana is a little (w an b wever info n proc written ers he causal in whi the give mied by one mo of rec mac is found n a me on 191 )The weigh logy argu earch,and in favo Gie..al are cons why doc this?Computer programs I search and paral The rea on why human intormation processing uses a parallel searc and not a serial search probably lies in the relatively slow nera respons time(acting against serial search).This constraint does not apply to the fast processing of computers.Thus,cognitive psychology may be sufficient to tell us the structure of information processing but it may not answer deeper questions about why information processing should be configured in that particular way.The biological constraints imposed by the brain shape the nature and limitations of cognition. DOES NEUROSCIENCE NEED COGNITIVE PSYCHOLOGY? It would be no exaggeration to say that the advent of techniques such as functional imaging has revolutionized the brain sciences.For example consider some of the newspaper headlines that have appeared over the years (Figure 1.8).Of course,it has been well known since the nineteenth century that pain,mood,intelligence and sexual desire are largely products of processes in the brain.The reason headlines such as these are extraordinary is because now the technology exists to be able to study these processes in vivo. Of course,when one looks inside the brain one does not "see"memories, thoughts,perceptions and so on (i.e..the stuff of cognitive psychology). Instead,what one sees is gray matter,white matter,blood vessels and so on (i.e.,the stuff of neuroscience).It is the latter,not the former,that one observes when conducting a functional imaging experiment.Developing a
12 THE STUDENT’S GUIDE TO COGNITIVE NEUROSCIENCE speakers), presenting the words in different locations on the screen and so on. This would provide further insights into the nature of this mechanism (e.g., what aspects of vision does it entail? Does it depend on word meaning?). However, both reaction-time measures and brain-based measures could be potentially informative. It is not the case that functional imaging is merely telling us where cognition is happening and not how it is happening. Another distinction that has been used to contrast cognitive psychology and cognitive neuroscience is that between software and hardware, respectively (Coltheart, 2004b; Harley, 2004). This derives from the familiar computer analogy in which one can, supposedly, learn about information processing (software) without knowing about the brain (hardware). As has been shown, to some extent this is true. But the computer analogy is a little misleading. Computer software is written by computer programmers (who, incidentally, have human brains). However, information processing is not written by some third person and then inscribed into the brain. Rather, the brain provides causal constraints on the nature of information processing. This is not analogous to the computer domain in which the link between software and hardware is arbitrarily determined by a computer programmer. To give a simple example, one model of visual word recognition suggests that words are recognized by searching words in a mental dictionary one by one until a match is found (Forster, 1976). The weight of evidence from cognitive psychology argues against this serial search, and in favor of words being searched in parallel (i.e., all candidate words are considered at the same time). But why does human cognition work like this? Computer programs can be made to recognize words adequately with both serial search and parallel search. The reason why human information processing uses a parallel search and not a serial search probably lies in the relatively slow neural response time (acting against serial search). This constraint does not apply to the fast processing of computers. Thus, cognitive psychology may be sufficient to tell us the structure of information processing but it may not answer deeper questions about why information processing should be configured in that particular way. The biological constraints imposed by the brain shape the nature and limitations of cognition. DOES NEUROSCIENCE NEED COGNITIVE PSYCHOLOGY? It would be no exaggeration to say that the advent of techniques such as functional imaging has revolutionized the brain sciences. For example, consider some of the newspaper headlines that have appeared over the years (Figure 1.8). Of course, it has been well known since the nineteenth century that pain, mood, intelligence and sexual desire are largely products of processes in the brain. The reason headlines such as these are extraordinary is because now the technology exists to be able to study these processes in vivo. Of course, when one looks inside the brain one does not “see” memories, thoughts, perceptions and so on (i.e., the stuff of cognitive psychology). Instead, what one sees is gray matter, white matter, blood vessels and so on (i.e., the stuff of neuroscience). It is the latter, not the former, that one observes when conducting a functional imaging experiment. Developing a
INTRODUCING COGNITIVE NEUROSCIENCE 13 Brain to blame for teenage mood swings Science uncovers apes'hidden'soul' Sex:it's all in the mind rswil map the di love and lust by peering into the male skull FIGURE1.8:The media framework for linking the two will necessarily entail dealing with the mind- loves to simplify the findings body problem either tacitly or explicitly.This is a daunting challenge. oMgcoentweaerosCei Is functionalimaging going to lead to a more sophisticated understanding appear to regard it as of the mind and brain than was achieved by the phrenologists?Some of the newspaper reports in Figure 1.8 suggest it might not.One reason why counterintuitive that sex phrenology failed is because the method had no real scientific grounding:the same cannot be said of functional imaging.Another reason why phrenology failed was that the psychological concepts used were naive.It is for this reason that functional imaging and other advances in neuroscience do require the insights from cognitive psychology to frame appropriate research questions and avoid becom ng a new phrenology (Uttal,2001). The question of whether cognitive,mind-based concepts will eventually become redundant (under a reductionist account)or coexist with neural- based accounts (eg.. as in dual-aspect theory)is for the future to decide. But for now,cognitive,mind-based concepts have an essential role to play in cognitive euro science FROM MODULES TO NETWORKS What does the future of cognitive neuroscience look like?Although nobody knows for sure,much current research is centered on understanding the mind and brain as a network.A network is a dynamically changing pattern of activity over several brain regions.Rather than thinking of the brain as a single network,there might be a multitude of different networks which are, themselves,switched on or off depending on the kind of thought or behavior KEY TERM that is needed.Thus,not only do brain regions have a degree of functional Connectome specialization,but entire networks may also have some specializations. A comprehensive map This network approach is exemplified by current efforts to map the human connectome(Sporns,2011).The Human Connectome Project was launched in 2010 at a cost of over S40M.The aim is to try to map out the pattern thought of as its"wiring diagram." of connectivity in the human brain at a macro.i.e..millimeter,scale (rather
Introducing cognitive neuroscience 13 framework for linking the two will necessarily entail dealing with the mind– body problem either tacitly or explicitly. This is a daunting challenge. Is functional imaging going to lead to a more sophisticated understanding of the mind and brain than was achieved by the phrenologists? Some of the newspaper reports in Figure 1.8 suggest it might not. One reason why phrenology failed is because the method had no real scientific grounding; the same cannot be said of functional imaging. Another reason why phrenology failed was that the psychological concepts used were naïve. It is for this reason that functional imaging and other advances in neuroscience do require the insights from cognitive psychology to frame appropriate research questions and avoid becoming a new phrenology (Uttal, 2001). The question of whether cognitive, mind-based concepts will eventually become redundant (under a reductionist account) or coexist with neuralbased accounts (e.g., as in dual-aspect theory) is for the future to decide. But for now, cognitive, mind-based concepts have an essential role to play in cognitive neuroscience. FROM MODULES TO NETWORKS What does the future of cognitive neuroscience look like? Although nobody knows for sure, much current research is centered on understanding the mind and brain as a network. A network is a dynamically changing pattern of activity over several brain regions. Rather than thinking of the brain as a single network, there might be a multitude of different networks which are, themselves, switched on or off depending on the kind of thought or behavior that is needed. Thus, not only do brain regions have a degree of functional specialization, but entire networks may also have some specializations. This network approach is exemplified by current efforts to map the human connectome (Sporns, 2011). The Human Connectome Project was launched in 2010 at a cost of over $40M. The aim is to try to map out the pattern of connectivity in the human brain at a macro, i.e., millimeter, scale (rather FIGURE 1.8: The media loves to simplify the findings of cognitive neuroscience. Many newspaper stories appear to regard it as counterintuitive that sex, pain and mood would be products of the brain. Sunday Times, 21 November 1999; Metro, 5 January 2001; The Observer, 12 March 2000. KEY TERM Connectome A comprehensive map of neural connections in the brain that may be thought of as its “wiring diagram