Full text PDF - Quantitative Methods for Psychology
Contents
1. amp Ullsperger P 1992 Mismatch negativity in event related potentials to auditory stimuli as a function of varying interstimulus interval Psychophysiology 29 546 550 Cowan N Winkler I Teder W amp Naatanen R 1993 Memory prerequisites of mismatch negativity in the auditory event related potential ERP Journal of Experimental Psychology Learning Memory and Cognition 19 909 921 Escera C amp Grau C 1996 Short term replicability of the mismatch negativity Electroencephalography and Clinical Neurophysiology 100 549 554 Girard M H Perrin F Pernier J amp Bouchet P 1990 Brain generators implicated in the processing of auditory stimulus deviance A topographic event related potential study Psychophysiology 27 627 640 McGee T Kraus N amp Nicol T 1997 mismatch negativity An assessment of methods for Is it really a determining response validity in individual subjects Electroencephalography and Clinical Neurophysiology 104 359 368 Naatanen R 1995 The mismatch negativity a powerful tool for cognitive neuroscience Ear and Hearing 16 6 18 N t nen R amp Alho K 1995 Mismatch negativity A unique measure of sensory processing in audition International Journal of Neuroscience 80 317 337 Naatanen R Gaillard A W K amp Mantysalo S 1978 Early selective attention effect on evoked potential reinterpreted Acta Psycologica 42 3132. 22 Figure 8 The Run Macro dialog box Run Macro O Workfiles v fej ecunitchansOLD vabs ee ExportCoherence_Alpha vabs i ExportCoherence_Beta vabs f ExportCoherence_Delta vabs i ExportCoherence_Gamma vabs fa ExportCoherence_Theta vabs Regarder dans Hi initchans2 vabs fe LatenceFenetre 9 macro Latencet Hi Macro LatenceF fA Macro LatenceF fa PeakExport vab EGIInitChans_042004 i AlternatingSign26apr vabs f Complex2Real26apr vabs Complex Data Measures yabs ee EGlInitChans vabs lt gt Nom du fichier CompareNodes vabs Fichiers de type Analyzer Macros vabs l Annuler Figure 9 The CompareNodes dialog box CompareNodes Comparison file and node ActveF ile Standard B History file History node Use external comparnson match file Fite Comparison file Output options Subtract Comparison type to perform Subtract comparison node before calculation f OF Cancel The History file field defines the node from which subtraction will be performed Make sure to keep the default ActiveFile option as it reflects the previously selected deviant stimulus node The History node field contains what To select the appropriate History node use the arrow and scroll down In the Comparison type to perform field of the Output options section will be subtracted from the Active File until you find the desired standard stimul
3. 329 Pekkone E Jousmaki V Partanen J amp Karhu J 1993 Mismatch negativity area and age related auditory memory Electroencephalography and Clinical Neurophysiology 87 321 325 Pekkonen E Rinne T amp N t nen R 1995 Variability and replicability of the mismatch negativity Electroencephalography and Clinical Neurophysiology 96 546 554 Ponton C W Don M Eggermont J I amp Kwong B 1997 Integrated mismatch negativity MMNi A noise free representation of evoked responses allowing single point distribution free statistical tests Electroencephalography and Clinical Neurophysiology 104 381 382 Schr ger E amp Winkler I 1995 magnitude of stimulus deviance effects on human pre Presentation rate and attentive change detection Neuroscience Letters 193 185 188 Sharma A Kraus N McGee T Carrell T amp Nicol T 1993 speech as reflected by the mismatch negativity event Acoustic versus phonetic representation of related potential Electroencephalography and Clinical Neurophysiology 88 64 71 Tervaniemi M Lehtokoski A Sinkkonen J Virtanen J Ilmoniemi R J amp N t nen R 1999 Test retest reliability of mismatch negativity for duration frequency and intensity changes Clinical Neurophysiology 110 1388 1393 Appendix I Procedure to follow to obtain the MMN The procedure followed to extract the MMN component will be demon
4. al 1993 In adults the maximum amplitude of this negative deflection is obtained over the frontal and central regions of the scalp suggesting that its primary source is located in the supratemporal auditory 19 cortex Girard et al 1990 Moreover this MMN occurs roughly 200 ms after stimulus onset and necessitates 200 250 presentations of the deviant stimulus in order to obtain a reliable and consistent MMN waveform component McGee et al 1997 The MMN is usually computed as ERPs evoked by a standard stimulus are subtracted from ERPs evoked by the presentation of a deviant stimulus Fig 3 The most common variables examined when studying MMN are the amplitude in uV the latency from stimulus onset in ms and the area under the curve in uV ms of the peak of interest Figure 3 The MMN wave is a subtraction of the ERP to the standard stimulus from the ERP to the deviant stimulus Note that in this figure the polarity is inverted negative up r Law Pie MMN i Standard To illustrate the application of the dilemma that will be presented in section 4 the next section of this paper will describe our latest study using the MMN component Presentation of a MMN study Beauchemin De Beaumont Turcotte Arcand Vannasing Belin amp Lassonde 2005 The MMN has received substantial scientific attention in the last decades as it is thought to reflect a the activation of cerebral mechanisms essential to pre
5. kfsa al Display Montage Transformations Export History Temp MadimMe_v212 Raw Data Tnithans OcularCorrection New ReferenceMast RawDatalnspecto r Fitters1 30 Standard DCDetrend Artifact Rejection Baseline orrection H Standard B voixE DCDetrend Artifact Rejection z Baseline orrection FAY voixe B A MmnE fal PeaksMrnon An Area Information Export dialog box will then appear It allows you to export the area information of an interval Specify in the Input section that the calculation of the area under the curve is Time Domained You then have to enter the values of the interval of interest Area Interval Relative to Time 0 which you will obtain by subtracting and adding by hand 25 ms to the value found by the peak detection procedure that has just been performed You also are requested to enter the Name of the Involved Data Sets the peak 24 node that you have selected just before requesting the Area Information Export window By checking the Primary History Files Only the selection will be confined to primary history files only It is important to select the Individual History Files option as you do not want to include all files in the workspace Remember that the interval in which the area under the curve will be measured must be calculated individually for each participant and each deviant stimulus peak detection In the Available Files section you will se
6. 2 MaQ6FD_y212 MaQ6FD_v412 Ma0 GS_v212 MaQ GS_v412 MaQ84P_yv212 MalOBB_v212 Add All gt gt Mal1LDB_v212 MalalP_v212 Mal3C4_v212 lt lt Remove All Mal4SC_v212 Mal5MD_v212 Received July 12 2005 Accepted September 22 2005
7. Tutorials in Quantitative Methods for Psychology 2005 Vol 1 1 p 18 24 Statistical analysis of the mismatch negativity To a dilemma an answer Maude Beauchemin 2 Louis De Beaumont 1Universit de Montr al Hopital Ste Justine This paper offers a new innovative outlook on mismatch negativity MMN analysis Indeed researchers in this field encounter difficulties when attempting to objectively quantify the MMN component waveform Advantages taken from already existing amplitude and area under the curve measures were used in order to thwart weaknesses from each individual measure The present paper can also be used as a guideline that describes each step required in the execution of the proposed technique to MMN analysis Ce travail sugg re une nouvelle approche a l analyse de la MMN En effet certains probl mes sont engendr s par les outils couramment utilis s pour analyser la MMN notamment l amplitude et l aire sous la courbe La technique sugg r e afin de d velopper une mesure objective de la MMN propose d utiliser les forces des deux techniques pr c demment nomm es afin de pallier a leurs faiblesses respectives Le pr sent travail se veut galement un mode d emploi quant a la fa on d appliquer les tapes n cessaires a la r alisation de cette nouvelle approche a l analyse de la MMN We first wanted to do a tutorial about the BrainVision Analyser program which processes raw EEG
8. a rs 1 1 a P A A l W Kee Thea Rhy er me din Pg a al gear Fin keat za i Th a ee a EO m ee sa mp P whatiia Wiis a le ed T a 1 rh se Pyl ia M ela Poy Hh lh Wark r lai y pl al ae int alt pattie fren Ma petit elt hata ruber a cra i if re al A a Fi Par fai Fi Fall Pat aT ed ra ral a awa aera a he i ay ged ne ey la heet erai a A cee i e a Awe A oF hans i il gh a en Me fe ae A npa et aja am a eal E af 5 ae ain a a Ee fn Tie rye pe a N Fh aay a api a a ee ee y mE a a ee am F 4 es ah F 7 i r a Le aa i aie TT ag TAD T Am FE ho ee eee i Tia A p m pe a yira M s yee ace io ari ie Yet A ay Loh ng u Phen a Pu I AA a ae TR er a ee ph i hras d E a z ij al i y ae y T fia apa i ee ei seek fe a eS e an ia a a ee e i ian ad ae ee If a stimulus event such as a sound is presented some of the measured neural activity will reflect the processing of that sound event This activity is termed the event related potential ERP However on a single trial the neural activity unrelated to the sound event which is usually referred to as noise typically precludes observation of the ERP waveform of interest Thus many trials of the sound event must be administered The resulting waveforms are afterwards lined up according to the onset of the sound events and then averaged Fig 2 Figure 2 Once averaged this is what an ERP wave looks like Note the ERP components e g P1 posi
9. attentive auditory discrimination and b the echoic sensory memory that underlies the latter discrimination process This growing interest toward the MMN has originated as considerable efforts have been exerted to disclose an objective measure of primary auditory information processing capacities The current study sought to determine whether the MMN could be used as an objective measure of voice familiarity More specifically this study tempted to verify whether the evoked MMN response was of greater amplitude when the deviant stimulus is a familiar voice as opposed to an unfamiliar voice as it may suggest that pre attentive mechanisms are implicated in voice recognition The main result of the present study is the significant difference between the MMN area under the curve elicited by a familiar voice when compared to that of an unfamiliar voice Fig 4 Figure 4 Grand average MMNs elicited by a familiar voice dashed and by an unfamiliar voice dotted referenced to the standard stimulus ERPs black The dilemma Before considering looking at variables of interest amplitude latency area under the curve stated in section 2 ERP data must first be filtered and analysed using a computer program such as BrainVision Analyser refer to the Vision Analyser User Manual version 1 05 Brain Products GmbH 1999 2004 for more detailed information This program enables users to extract the MMN component waveform by scripting a macr
10. ce Thus latency computations will not be the focus of this paper as a fairly straightforward set of 20 operations is sufficient to obtain this variable using the BrainVision Analyser program refer to the Vision Analyser User Manual version 1 05 Brain Products GmbH 1999 2004 Another variable that has traditionally been extracted when analysing the MMN component is its amplitude using averages Over various time intervals such as an interval around the peak latency The amplitude of the MMN generally increases as the difference between the standard and the deviant stimuli is enhanced This relationship is generally monotonic although it tends to level off as the difference between the standard and the deviant stimuli becomes large Schroger amp Winkler 1995 Therefore the amplitude should not be utilized to quantify the MMN Other investigators seem to prefer reporting the area under the curve to account for the size of MMN activation Pekkonen et al 1993 Sharma et al 1993 McGee et al 1997 However the duration of the temporal window in which the MMN waveform occurs varied fairly across participants Figures 5 and 6 Should one consider the peak amplitude to be the most indicative variable to reflect brain activation or is the area under the curve contained within a predefined temporal window more appropriate How can experimenters account for such variability in the MMN waveform configuration Figure 5 MMN elici
11. data both for spontaneous EEG analyses and for evoked potentials However its user friendly workspace designed to allow users to interactively compute complex analysis tasks combined with the already existing comprehensive Vision Analyser User Manual version 1 05 Brain Products GmbH 1999 2004 which contains detailed information on how to design a multi step analysis have changed our plans In fact we did not want this paper to be a replicate of what had already been made accessible to the public Instead we have decided to propose a new perspective on Mismatch Negativity MMN analysis with a brief introduction on what the MMN component This paper begins stands for its origins and its associated variables of interest We will then present a dilemma frequently encountered by researchers conducting ERP studies using the MMN component and provide you with what we consider the most appropriate way to resolve the issue 18 EEG ERP and MMN The voltage difference between an electrode placed at a position of interest on the scalp and a reference electrode placed at a relatively neutral position with respect to the neural activity of interest yield an electroencephalogram EEG More specifically the EEG is a time varying voltage signal that reflects the activity of many neurons working in concert Fig 1 Figure 1 An oe trace PLE am Pi 7 L F asl Tar Ai p Nes yl j f gg i es ain al f 7 a va px il ae olf Amit
12. e all of your files select the appropriate one by clicking the Add button The selected file will then be transferred into the Selected Files section Make sure to Use Activity to rectify the sign so the values are unsigned and to select the Area Export Type Finally name your output in the Output File field of the Output section You have to repeat all the above steps for each participant for both deviant stimuli Once you have completed those steps you can open the output files in a Microsoft Excel sheet Organize you data so as to facilitate statistical analyses performed with SPSS or any other statistical program of your choice Figure 16 The Area Information Export dialog box Area Information Export Input Options Time Domain Frequency Domain F Overwrite Default Decimal Symbol Area Interval Relative to Time 0 E Start ms 1175 End ms 1225 PEREAT Name s of the Involved Data Sets History Nodes Separated by Commas Use Activity Unsigned Values Rectified PeakMmnE C Use Voltage Signed Values Export Type Area yy ms C Mean Activity p Select Individual History Files is Selection Filter Refresh Available Files V Primary History Files Only C Use Whole Workspace Output Selected Files Folder C DATASMMN Voix HistoryFiles MaQIMB_v412 MaQiMB_v212 Ma02VT_v212 Output File Area50_Ma01MB_v21q tet Ma034T_v212 Ma034T_v412 Ma04DT_v212 MaQ4DT_v412 Ma05GO0_v212 Ma05G0_v41
13. e local maxima detection method which would exclude these edge points Figure 13 Peak Detection dialog box Step 2 of 3 Peak Detection Step 2 of a Prak Tahle interead Hane Stal fee Endis Pokal Coku Incest Lin m e al E ec Fama Lins al Harrera Jul H E is a E a A u A z Do E E Ez 7 H i 4 Prbnideni Arida On the second page of the dialog box you have to name the peak that you are searching for indicate the window in which the peak must be searched in as well as the polarity of the peak On the third pane you must enable channels of interest and click on Finish when this is done Go through step one to three for the other deviant stimulus Figure 14 Peak Detection dialog box Step 3 of 3 Peak Detention Step daf d Channels Peate enable She channels in steh She peaks should be marked gt e Dibi lt Pr c dent Teren Furies Now that the most negative point of the MMN has been peaked you can now require BrainVision to calculate the area under the curve as follows First select the Peak node of the MMN for one deviant stimulus Then click on Export gt Area Information Export Figure 15 The Peak node newly created is selected in order to perform the area under the curve calculation Analyzer Raw Data lnitChans OcularCorrection New Referer File Edit view Djek Bs S ol ls
14. it wiew Display Montage Transformations Export History Temp Oela Ba al ajals efs F Madime_v212 Raw Data InitChans OcularCorrection New ReferenceMast RawDatalnspecto r Filters1 30 Standard DCDetrend Artifact Rejection BaselineCorrection A Standard 6 WoixE DCDetrend Artifact Rejection BaselineCorrection EA voixe B gt e 23 You click on Transformations gt Peak Detection to initiate a 3 step procedure for peak detection to occur Figure 12 Peak Detection dialog box Step 1 of 3 Peak Detection Step 1 of 3 Methods Auuboratcen Mahadi Searchong Haradi i Semsulomate Delecdam Separate Seach bo Even Charel Automatic Detector 0 Search Peak ina Aeleence Channel and Se Makart wrth Pledoect bo this Pesk Delechon Mellnals E Sanch ior Glishal Maina in niera Seach bor Local Haaman Intenral Search bor Weeghted Local Mirana m interval iunie This firs step allows you to control for the degree of automation With the semiautomatic detection a cursor will be positioned where the algorithm detected the peak In the Searching Methods section you have to choose the option Separate Search for every channel You also have to choose the desired search method for peaks in the Detection Methods section In searching for a global maxima the edge points of the interval will be included when looking for peaks within the interval rather than th
15. o a specific program specifically designed to subtract ERPs evoked by a standard stimulus from ERPs evoked by the presentation of a deviant stimulus to access the step by step method of calculation of the MMN refer to Annex I Once the MMN waves are obtained we can then consider examining the variables of interest as statistical quantification of the MMN The latency of the MMN can be interpreted as the time required to distinguish a deviant stimulus from a standard analysis requires optimized stimulus In terms of sensory discriminations the difference in timing when processing different stimuli is thought to account for the discrimination of subtle differences between the presented stimuli When interpreting the latency of the MMN it is important to disentangle the level of difficulty of the discrimination task from the timing of the discrimination process In fact if one wishes to determine which type of discrimination occurs earlier in the auditory system it is essential to control for discrimination task difficulty when measuring MMN latencies Applied to the above example when comparing MMN elicited by a familiar voice to that of an unfamiliar voice if the latencies of the two MMN components were found to be different it would suggest that one voice is analysed prior to the other As illustrated in Figure 4 no latency differences were found between the MMN elicited by a familiar voice with that elicited by an unfamiliar voi
16. strated using data from the applied example Firstly you must filter ERP data for artefacts and ocular corrections refer to Vision Analyser User Manual version 1 05 Brain Products GmbH 1999 2004 have been performed the Once these MMN component will be obtained using a pre programmed preliminary steps macro which requires an active history node for operation make sure you selected the deviant stimulus node before activating the macro This macro is programmed so as to subtract the ERP elicited by a frequent stimulus from that of a deviant stimulus Figure 7 Note that all the filtering and analysing steps are done The VoixE B is one of the two deviant stimuli Il is selected so as to the macro File Edit View Display Montage Transformations Export History Temp Oelk e S a aale ENA MaoiMB_v212 Raw Data TnitChans CcularCorrection New ReferenceMast RawDatalnspecto r Filters1 30 Standard DC Detrend Artifact Rejection BaselineCorrection A Standard B WoixE DC Detrend Artifact Rejection BaselineCorrection Then you click on Macro gt Run to obtain the Run Macro dialog box Select the CompareNodes program which will allow you to perform the subtraction requested for the computation of the MMN A second dialog box specific to CompareNodes will then appear In this dialog box you will need to specify the history nodes you intend to subtract from another
17. ted by the same deviant stimulus for two different pariticipants at the same electrode site Note that although both have the same amplitude one is wider than the other one Figure 6 MMN elicited by the same deviant stimulus for two different participants at the same electrode site Note here that although both have the same width their amplitudes differ A solution This section will attempt to describe what we consider to be the most appropriate way to resolve the high variability in MMN waveform configuration to account for brain activation Thus in order to develop an objective way to quantify the curve of a MMN component advantages from both approaches amplitude and area under the curve were used Therefore as used in the applied example the MMN component was obtained using the area under the curve contained within a 50 ms time window in which the midpoint had previously been identified in a peak amplitude detection manipulation performed for each participant The MMN component values obtained when presented with an unfamiliar voice were then compared to those elicited by a familiar voice This method appears to be optimal since using a predefined window in which to compute the area under the curve prevents investigators to insert potential bias by the variability of the duration of the temporal window in which the MMN waveform occurs Moreover centering this defined window around the peak amplitude accounts for the grea
18. test difference between the standard and deviant stimuli Appendix II provides a detailed description on how to use BrainVision Analyser to obtain the area under the curve values Conclusion Although the MMN has been useful in furthering scientific knowledge about auditory processing its use is limited by the Nevertheless considerably interindividual response variability studies are systematically addressing the issue of test retest reliability Escera amp Grau 1996 Tervaniemi et al 1999 Pekkonen et al 1995 while other groups are actively looking at ways to enhance the objective quantification of the elicited MMN response Ponton et al 1997 McGee et al 1997 The present paper also intended to propose a new perspective on MMN analysis Despite recent methodological advances that enabled investigators to reduce interindividual variability it remains unclear whether the MMN is sufficiently reliable to be used in clinical settings However the MMN is one of the few biological indexes of fine tuned sensory perception as it will most likely continue to yield significant insights about the processing of auditory information in various research and clinical endeavours References Beauchemin M De Beaumont L Turcotte A Arcand C submitted of voice Vannasing P amp Belin P Lassonde M MMN an familiarity European Journal of Neuroscience electrophysiological marker 21 Bottcher Gandor C
19. tive component peaking at about 100 ms a00 00 500 05 a i 200 Time ms The MMN is a change specific component of the auditory event related potentials ERPs Indeed the MMN is a versatile measure that can discriminate the smallest alterations when any one parameter differs between two consecutive stimuli Auditory oddball paradigms which involve the presentation of infrequent stimuli embedded among frequent stimuli have commonly been used to generate the ERP component called MMN Naatanen et al 1978 Naatanen amp Alho 1995 According to Naatanen and Alho s 1995 model the discrimination of two successive stimuli differing in only one parameter reflects the involvement of two different neural representations In other words a frequently presented stimulus forms a neural trace in the echoic sensory memory which can last up to 8 to10 seconds Bottcher Gandor amp Ullsperger 1992 The sensory input from the deviant stimulus does not fit with the existing neural trace therefore resulting in a negative deflection the MMN component Thus the MMN is elicited by any discriminable change in some repetitive aspects of auditory stimulation stored in echoic memory Importantly the MMN is not elicited by those deviant stimuli when standard stimuli are omitted N t nen 1995 implying that the MMN indexes the discrepancy between the incoming stimulus and the memory representation of the standard stimulus Cowan et
20. us select Subtract and then press OK Figure 10 The history tree showing the added node for the MMN of one deviant stimulus Analyzer Raw Data InitChans OcularCorrection Mew Referer E File Edit View Display Montage Transformations Export History Temp Delil Bal S O S 2fso ET al MadimMB_v212 Raw Data fA InitChans A OcularCorrection A New ReferenceMast A RawDatalnspecko r E Fiters1 30 A Standard E DC Detrend A Artifact Rejection 4 A Baseline Correction A Standard B EA voixe A OCDetrend H Artifact Rejection A BaselineCorrection FAY voixe Fal Minne Note that an extra node that contains your extracted MMN was added Repeat the above steps to obtain the MMN for the other deviant stimulus of interest Appendix II The new innovative area under the curve variable Once you have completed the steps in Appendix I a peak detection procedure is then executed to identify the most negative deflection in a literature based predefined time window between 80 280 ms In order to perform this peak detection step you must first select the MMN node obtained after you have accomplished the Appendix I procedure of one deviant stimulus in the applied example either the familiar or the unfamiliar voice Figure 11 The MMN node obtained after you have accomplished the Appendix I is selected Analyzer Raw Data InitChans OcularCorrection New Referer E File Ed
Download Pdf Manuals
Related Search