Treffer: A Free Software Package for Implementing Psychological and Psychophysiological Experiments
Ghent University, Belgium
CC BY 4.0
Sauf mention contraire ci-dessus, le contenu de cette notice bibliographique peut être utilisé dans le cadre d’une licence CC BY 4.0 Inist-CNRS / Unless otherwise stated above, the content of this bibliographic record may be used under a CC BY 4.0 licence by Inist-CNRS / A menos que se haya señalado antes, el contenido de este registro bibliográfico puede ser utilizado al amparo de una licencia CC BY 4.0 Inist-CNRS
FRANCIS
Weitere Informationen
We describe Affect 4.0, a user-friendly software package for implementing psychological and psychophysiological experiments. Affect 4.0 can be used to present visual, acoustic, and/or tactile stimuli in highly complex (i.e., semirandomized and response-contingent) sequences. Affect 4.0 is capable of registering response latencies and analog behavioral input with millisecond accuracy. Affect 4.0 is available free of charge.
Affect 4.0: A Free Software Package for Implementing Psychological and Psychophysiological Experiments
<cn> <bold>By: Adriaan Spruyt</bold>>
>
> <bold>Jeroen Clarysse</bold>
>
> <bold>Debora Vansteenwegen</bold>
>
> <bold>Frank Baeyens</bold>
>
> <bold>Dirk Hermans</bold>
>
<bold>Acknowledgement: </bold>Preparation of this paper was supported by GrantBOF/GOA2006/001 of Ghent University. Adriaan Spruyt is nowPostdoctoral Fellow of the Flemish Research Foundation (FWO –Vlaanderen) and Research Fellow at the University of Leuven, Belgium. JeroenClarysse, Debora Vansteenwegen, Frank Baeyens, Dirk Hermans, University ofLeuven, Belgium. The development of Affect 4.0 was supported by GrantsGOA/2001/01 and GOA/2007/03 of theUniversity of Leuven.
The personal computer has revolutionized the face of experimental psychology: It allows for flexible, well-controlled stimulus presentations across various experimental paradigms as well as the registration of a wide range of behavioral responses with high temporal accuracy. Yet, these obvious advantages come at a cost of an ever-growing amount of technical knowledge that is required to develop experimental programs in a valid manner. To reduce the programming skills required to implement computer experiments, researchers often rely on software packages that comprise a graphical user interface (GUI) and/or simple syntax. Well-known examples are Direct RT v2008, E-Prime 2.0, Inquisit 3.0, and SuperLab 4.0 (for a recent review, see Stahl, 2006). Today, these software packages are widely used in both research and educational settings, and they certainly contribute to the elevation of common experimental praxis to a state-of-the-art level. However, it may be problematic for at least a number of potential users that each of these software packages is distributed commercially at a substantial cost (see Stahl, 2006). In this article, we describe Affect 4.0, a free alternative to commercial software packages that was developed at the Centre for the Psychology of Learning and Experimental Psychopathology of the Catholic University of Leuven (Belgium).
The motivation to develop Affect 4.0 was fourfold. First of all, we found the implementation of complex semirandomizations of different stimuli across different trial types to be quite difficult and/or laborious in existing software packages (if possible at all). Second, we frequently experienced difficulties interfacing between the computer controlling an experiment and external hardware that is typically used in psychophysiological setups. Third, even though existing software packages are relatively easy to use, they still require programming skills that exceed the proficiency of a typical graduate student. Finally, we found the options to implement response-contingent stimulus presentations (e.g., error feedback) to be fairly limited in existing software packages. Affect 4.0 deals with each of these concerns. Thanks to the GUI of Affect 4.0 and the possibility to use “wizards”, Affect 4.0 enables researchers and students alike to implement highly complex semirandomizations and experimental procedures in an intuitive and straightforward software environment. In this article, we will highlight the critical characteristics of Affect 4.0. It is not our intention, however, to present a detailed user manual. For the Affect 4.0 download, manual, tutorial, help files, and sample experiments we refer the reader to the Affect 4.0 website at
Overview
>
We will first describe the Affect 4.0 GUI and demonstrate how typical “building blocks” of an experiment such as trials, output files, and experimental stimulus lists can be instantiated and manipulated. In a second section, we will describe the advantages of using Affect 4.0 for psychophysiological research. Finally, the timing accuracy of Affect 4.0 and similar software packages will be discussed.
The Affect 4.0 GUI and Object Classes
>
Affect 4.0 stores .af4 documents in plain text, structured as eXtensible Markup Language (XML). The GUI of Affect 4.0 builds itself dynamically on top of these XML-structured documents. This means that one can use any XML or text editor to edit .af4 documents, although we recommend inexperienced users to rely on the Affect 4.0 GUI. The Affect 4.0 GUI itself is hierarchically structured (see Figure 1). At the highest level, it presents an overview of all possible “object classes”. Within each object class, new objects can be created, duplicated, or deleted by a simple mouse click, and each of these objects can be selected for further editing at a lower level. The order in which the Affect 4.0 object classes are listed in the Affect 4.0 GUI corresponds to the order in which they should be created and edited by the user. In what follows, we will briefly describe the Affect 4.0 object classes. For convenience, generic Affect 4.0 concepts will be printed in a special font (e.g., flipbook). Concrete examples of user-defined exemplars of such concepts will be written between quotation marks (e.g., “name_of_a_flipbook”).
>
><anchor name="fig1"></anchor>
Stimuli are static objects that do not require complex randomizations across different trial types. Typical examples are fixation points and feedback messages. Eight types of stimuli can be defined: Pictures (bitmaps only), character strings (e.g., words and/or nonsense words), sounds (DirectX compatible, see the DirectX Developer Center at the Microsoft Developer Network), flipbooks, mosaics, Labmaster (digital) output signals, National Instruments output signals (digital and/or analog), and parallel port output signals (TTL, transistor-transistor logic).
A flipbook is a bitmap consisting of a series of frames that together create the illusion of movement when they are presented in quick succession. Consider, for example, the stimulus-response compatibility (SRC) task of Bradley, Field, Mogg, and De Houwer (2004; see also Mogg, Bradley, Field, & De Houwer, 2003). In this task, participants are required to move a manikin that is displayed in the upper or lower half of a computer screen by pressing one of two keys depending on the nature of a centrally presented target stimulus. In one half of the experiment, participants are instructed to approach one type of stimuli (e.g., smoking-related scenes) and to avoid stimuli of a second category (e.g., nonsmoking-related scenes). In the other half of the experiment, the stimulus-response assignment is reversed. The illusion of a walking manikin can be created by presenting different frames of the same flipbook in quick succession. The exact presentation duration of each frame can either be controlled by the experimenter (i.e., fixed presentation duration) or depend upon the behavior of a participant. Figure 2 illustrates how flipbooks can be used to create the illusion of a moving manikin.
>
><anchor name="fig2"></anchor>
A mosaic is a grid of visual stimuli (pictures and/or character strings), the content of which is based on a user-defined stimulus pool (i.e., a set or subset of experimental stimuli, see below). The location of each stimulus within a mosaic is random, but researchers can choose whether a different randomization is used each time the mosaic is presented or the same randomization is used time and again. Because the content of a stimulus pool can be semirandomly controlled by an experimenter (see below), the (visual) stimuli that appear in a mosaic can vary from one participant to another and even from one block of trials to another. Nevertheless, Affect 4.0 handles mosaics as static objects as it is not possible (a) to randomize their content on a trialwise basis or (b) to randomize different mosaics across different trial types. Note, however, that stimuli that are randomized on a trialwise basis across different trial types can be presented at random locations of a mosaic. For example, one could define a mosaic consisting of
Finally, stimuli can be digital or analog output signals that are generated by the parallel port (digital output signal, TTL), Labmaster data acquisition (DAQ) cards (digital output signals only), and/or National Instruments DAQ cards (digital and/or analog output signals). These output signals can be used to control external hardware. Typical examples are the Peripheral Electrical Stimulation devices of Digitimer (e.g., DS5, DS7A, and DS7AH).
<h31 id="zea-57-1-36-d260e188">Stimulus Pools</h31>Stimulus pools are lists of stimuli (character strings, pictures, and/or sounds) that one wants to present in a randomized order. The content of these stimulus pools is based on tab-delimited text files (e.g., “stimuli.txt”; see Figure 3) that list all to-be-presented stimuli (first column) as well as stimulus properties that distinguish between different categories of stimuli (from the second column onwards). To create stimulus pools that consist of pictures and/or sounds, the first column of the source text files must contain a list of all relevant file names (e.g., picture1.bmp and sound1.wav) and the corresponding picture and sound files need to be stored in the same folder as the source text file. For stimulus pools consisting of character strings, Affect 4.0 will simply use the strings of the first column as stimuli (case-sensitive). Figure 3 illustrates how a source text files might look like. In the first column, a set of 20 stimuli are listed. As indicated by the first stimulus property (second column), 10 of these stimuli are words whereas the other stimuli are pictures. The second stimulus property (third column) specifies the valence of each stimulus.
>
><anchor name="fig3"></anchor>
As an example, consider the emotional Stroop task (see Pratto & John, 1991; Williams, Mathews, & MacLeod, 1996). In this task, participants are asked to name the color of emotionally relevant words that appear one by one on the computer screen. To be able to present these words in a randomized order, it is sufficient to create a single stimulus pool containing all to-be-presented words. If one wants to present more than one randomly selected stimulus during the same trial, several stimulus pool are needed. For example, to implement a typical priming study (see Neely, 1991), an experimenter would create two stimulus pools, one for the primes and one for the targets. It is also possible to semirandomly divide to-be-presented stimuli across two or more virtual stimulus pools. For example, if all stimuli of a priming experiment are listed in a single source text file and one of the stimulus properties distinguishes between primes and targets, two virtual stimulus pools can be created (one containing all prime stimuli and one containing all target stimuli). These two virtual stimulus pools can then be used as if they were based on two different source text files. The same logic can also be applied to counterbalance specific stimuli across different stimulus pools for different participants or to control the content of mosaics (see above).
<h31 id="zea-57-1-36-d260e244">Output Files</h31>One can define several output files for the same experiment (e.g., one for all data of a practice phase and one for all data of an experimental phase), each of which can contain up to 128 data columns. Depending on the preferences of the experimenter, all output files of a single participant can be stored in a separate folder (i.e., multilevel output) or all output files of all participants can be stored in a single folder.
Affect 4.0 distinguishes between output files for analog behavioral input (e.g., skin conductance), on the one hand, and output files for discrete behavioral input on the other hand (e.g., response latencies). Because (a) analog behavioral measures require a specific structuring of the output files (i.e., one line per millisecond instead of one line per trial) and (b) different output files are required for each channel of analog behavioral input (e.g., one output file for skin conductance and one output file for startle responses), output files for analog behavioral input are implemented at the response level. We will describe these output files in the Psychophysiology section. In this section, we describe output files for discrete behavioral input only.
Discrete output files are tab-delimited text files and are highly customizable: It is possible to save whatever information (e.g., response latencies and customizable text) depending on any event. Affect 4.0 is also capable of evaluating user-defined expressions at runtime to determine the correctness of any response. This information can be saved in any particular data column. If the include default columns option is selected, Affect 4.0 will automatically add several data columns containing information about the characteristics of each trial. These default data columns comprise the Affect 4.0 version number, a timestamp, trial numbers, the block numbers, the trial properties (see below), a listing of all stimuli that are presented during a given trial as well as details about the timing accuracy (see below).
<h31 id="zea-57-1-36-d260e270">Responses</h31>At runtime, Affect 4.0 can monitor up to 512 events simultaneously. These events can be predefined moments in a trial flow or behavioral input (see below). Typical response devices that can be monitored are standard computer keyboards, computer mice, as well as game port and parallel port devices (e.g., Reacsys R-51 Voice Key). To allow monitoring of analog behavioral input (e.g., skin conductance and/or startle responses), Affect 4.0 can interface with Labmaster and National Instruments DAQ cards. Other DAQ cards are currently not supported. It is also possible to collect rating data using customizable rating scales (i.e., evaluations). However, because rating scales require one to present certain stimuli (e.g., the rating scale itself), they are not implemented at the response level of the Affect 4.0 GUI. Instead, they are implemented at the settings level.
<h31 id="zea-57-1-36-d260e280">Trials</h31>The exact sequence of events during any given trial (i.e., trial flows) as well as the nature of different types of trials (i.e., trial properties) can be defined in the trials object class. Six different types of flow events can be defined: Specific moments in a particular trial (i.e., at time “
To distinguish between trials that have identical trial flows but different stimulus characteristics, Affect 4.0 relies on user-defined trial properties. Typically, one will define several trial properties on the basis of individual stimulus properties and/or (complex) combinations of these stimulus properties. To illustrate the use of trial properties, consider a typical affective priming study in which participants are presented with primes and targets that either have a positive or negative connotation (e.g., Fazio, Sanbonmatsu, Powell, & Kardes, 1986). Provided that the source text files containing the prime and target stimuli also include a stimulus property that distinguishes between positive and negative stimuli (see Figure 3), one might define a trial property named “congruent_trials” as follows:
On the basis of this definition, Affect 4.0 will interpret any affectively congruent prime-target combination as such. Note that the order of the comma-separated feature specifications depends on the order in which stimulus pools were initially added to a trial definition. That is, if the stimulus pool containing the list of primes is added to the trial definition before the stimulus pool containing the list of targets, Affect 4.0 will treat the first position of each expression as referring to the primes and the second position as referring to the targets. Also note that the logic operands “AND” and “OR” can be used to define highly complex trial properties. For example, if word stimuli of an affective priming study were to be presented in blue or brown print (e.g., Klauer & Musch, 2002), one could define a trial property “both_dimensions_congruent” as follows:
As another example, imagine an experiment in which two primes are presented (see Gawronski, Deutsch, & Seidel, 2005) and all stimuli are randomly presented in 20 different colors instead of just 2 different colors. In this case, an exhaustive definition of all possible prime-target pairs that are congruent on both dimensions would require a list of 22 expressions. To reduce this complexity, Affect 4.0 can interpret definitions of trial properties that contain expressions in which the value of one stimulus depends on the value of other stimuli. For the present example, one might define the trial property “all_stimuli_congruent_on_both_dimensions” as follows:
The first expression specifies that the second prime and the target should always have the same color as the first prime (“=*1.color_first_primes”) while no restriction is specified for the first primes (hence the “*” at the location of the first prime in the definition). The second part of the definition specifies the additional restriction that the primes and the target should all have the same valence.
Crucially, because trial properties and trial flows are two distinct aspects of a trial definition, experimenters can first implement highly complex semirandomizations of the stimulus content and then execute a given trial flow using this semirandomized stimulus content. To that end, users first have to create one or more trial pools. Initially, these trial pools are simply empty sets, but they can be “filled” with any number of randomly selected stimuli or stimulus combinations while respecting restrictions that are based on user-defined trial properties. To that end, one or more fillers need to be defined in the fillers object class.<anchor name="b-fn2"></anchor><sups>2</sups>
<h31 id="zea-57-1-36-d260e452">Fillers</h31>The actual stimulus content that will be presented during an experiment is determined by means of one or more fillers, each of which can execute one or more filler commands (e.g., selecting a counterbalanced subset of trials). A complete overview of all possible fillers commands is provided at the Affect 4.0 website. Crucially, fillers will randomly select trial content (i.e., stimuli or combinations of stimuli) while respecting user-defined restrictions. For example, if one were to design a semantic priming experiment in which the prime and target stimuli represent animate or inanimate concepts (e.g., Klinger, Burton, & Pitts, 2000) and trial properties have been created for each of the four types of trial in the trial object class (i.e., animate_animate, animate_inanimate, inanimate_animate, inanimate_inanimate), equal amounts of randomly drawn prime-target combinations can be selected for each trial type by defining four property restrictions and specifying the desired number of trials. An example of such a filler definition can be found in Figure 4. Here it is illustrated how one may define a filler named “fill_block_1” to obtain a set of 200 prime-target combinations consisting of 100 congruent pairs (50 animate_animate, 50 inanimate_inanimate) and 100 incongruent pairs (50 inanimate_animate, 50 animate_inanimate).
>
><anchor name="fig4"></anchor>
Importantly, Affect 4.0 will avoid stimulus repetition by default: Unless specifically stated otherwise (as in Figure 4), no stimulus of a given stimulus pool will be presented more than once. Of course, if stimulus repetition is not an issue, it is still possible to reuse stimuli over and over again. Also, using multiple fillers and trial pools, the exact number of stimulus repetitions can be controlled. This is an important feature of Affect 4.0 as many experimental phenomena (e.g., negative priming, see Malley & Strayer, 1995; Strayer & Grison, 1999; Tipper, 1985) are known to depend upon stimulus repetition.
<h31 id="zea-57-1-36-d260e499">Experiments</h31>Once all required objects are defined, they can be combined in the experiment object class. For example, it is possible to run a given trial flow using the stimulus content of a given trial pool. In principle, multiple experiments can be defined using the same basic building blocks. At the experiment level, one can also specify instructions for participants, manipulate the inter trial interval (ITI), etc. An overview of all the experiment commands (e.g., running a filler and running a trial flow on a particular trial pool) is provided at the Affect 4.0 website.
Psychophysiology
>
Typically, presenting experimental stimuli and recording physiological measures simultaneously involves the use of two computers (e.g., Curtin, Lozano, & Allen, 2007). Whereas one computer is used to control the presentation of experimental stimuli, the other computer is used to register physiological measures such as skin-conductance responses, startle responses, heart rate, and respiration. To synchronize both computers, they are typically connected via parallel digital input-output (I/O) ports, and specialized stimulus-presentation software running on the first computer informs specialized data-acquisition software running on the second computer each time an important event occurs by sending an event marker.
While this technique certainly allows for accurate and reliable registration of physiological data, it also comes with important disadvantages. First of all, data files are inevitably bulky as the physiology-recording computer is continuously recording all signals during the whole experiment. Preprocessing of these data files can sometimes be time-consuming, especially for researchers who are unable to write or edit data-processing programs. Second, the number of event markers that one can send is fairly limited (eight-bit signals only) and require additional programming on the data-acquisition computer to be interpreted correctly by the data-acquisition software. Finally, it is almost impossible to implement dynamic experiments in which the stream of events is based upon a subject’s physiological data pattern (e.g., changing the background color of the computer monitor whenever the skin-conductance exceeds a certain threshold).
To deal with these issues, Affect 4.0 allows for the recording of physiological measures and the presentation of experimental stimuli on a single computer. Using direct memory access (DMA) features of DAQ cards, Affect 4.0 can now monitor up to three 1-kHz channels (e.g., skin conductance, startle responses, and online expectancy dial, Vansteenwegen, Iberico, Vervliet, Marescau, & Hermans, 2008), while presenting stimuli and reading keypad response latencies with millisecond accuracy. Note, however, that Affect 4.0 may occasionally fail to monitor physiological measures during the very first and/or very last millisecond of a given trial as a result of unavoidable round-off errors (for a discussion of timing accuracy under Windows, see below).
As already mentioned above, separate output files need to be created for each physiological measure. For that reason, the option to create output files for analog behavioral input is implemented at the level of the responses. In the first and second columns of these output files, Affect 4.0 will store the time (one line per millisecond) and the value of the analog behavioral measure, respectively. As of the third column, up to 64 different character string, flags, counters, and/or stopwatches can be added, allowing researchers to format their data files conforming to the requirements of existing data analysis software (e.g., PSPHA, De Clercq, Verschuere, De Vlieger, & Crombez, 2006). Affect 4.0 is also able to display graphs of physiological measures on a second monitor at runtime. That way, eventual problems with the physiological recordings can be detected very quickly and range or sensitivity adjustments can be performed if needed. If synchronization of stimulus presentation and MRI scanning is needed, Affect 4.0 can also monitor parallel port input signals (TTL). The possibility to implement dynamic experiments in which the stream of events is based upon a subject’s physiological data pattern is currently under investigation.
Timing Accuracy
>
Akin to software packages like E-prime, Inquisit, SuperLab, and DirectRT, Affect 4.0 was developed for the Windows Operating System (OS). Because Windows is not a realtime OS, some caution regarding timing accuracy is necessary (see also Myors, 1999). Whereas realtime OSs guarantee that running applications are never interrupted by the OS for a period longer than a predefined and fixed amount of time, programs running on the Windows OS can be halted indeterminately at any time to allow the OS to handle certain tasks. As a result, it is in principle impossible to guarantee millisecond accuracy at all times
In sum, although Windows is not a realtime OS, Affect 4.0 will rarely, if ever, fail to meet millisecond accuracy. In fact, Affect 4.0 will typically execute its core loop many times within one millisecond on modern machines. To verify whether a given computer supports millisecond accuracy, one can run Affect 4.0 experiment files in diagnostics mode. Affect 4.0 will then generate a detailed report of the timestamps of all time-critical events. Moreover, when defining the format of discrete output files (see above), users can choose to include default data columns (see above). Affect 4.0 will then keep a record (for each trial separately) of whether its core loop is completed within one millisecond at all times and store this information in one of these default data columns. Affect 4.0 also monitors whether or not all visual stimuli of a given trial fit into VRAM and saves this information in one of the default data columns. Note, however, that timing accuracy is also highly dependent upon the hardware that is used to register responses. Keyboards, for example, are inherently inaccurate due to keystroke buffering and preprocessing, overhead introduced in the PS2 and USB protocols, and the low priority of the windows keyboard drivers (e.g., Crosbie, 1990; Plant, Hammond, & Turner, 2004; Segalowitz & Graves, 1990; Voss, Leonhart, & Stahl, 2007).
Affect 4.0 in Practice
>
Affect 4.0 has been downloaded by researchers working at more than 50 different universities in more than 20 different countries. Over the past few years, Affect 4.0 and its earlier versions (Affect 1.0 and Affect 3.0) have been used in numerous experiments of different research groups. The results of those experiments have been published peer-reviewed articles on learning psychology (e.g., Vansteenwegen et al., 2007), neuropsychology and psychophysiology (e.g., Coppens et al., 2007), social cognition and attitudes (e.g., Spruyt, De Houwer, Hermans, & Eelen, 2007), experimental psychopathology (e.g., De Cort, Hermans, Spruyt, Griez, Schruers, 2008), implicit attitude measures (e.g., Spruyt, Hermans, De Houwer, Vandekerckhove, & Eelen, 2007), general experimental psychology (Spruyt, Hermans, De Houwer, & Eelen, 2004), and health psychology (e.g., Wan et al., 2008).
System Requirements, Download, and Installation
>
Affect 4.0 was designed for the Windows XP OS and can in principle be validly used on any modern computer (i.e., Pentium 4 @ 1 GHz or higher) with DirectX compatible graphics card with sufficient VRAM (see above). To ascertain that a computer is indeed capable of running Affect 4.0 with millisecond accuracy, we recommend users to verify the timing accuracy of their setup using the diagnostics mode of Affect 4.0 before using it in an experimental setting (see above). More technically skilled users can also rely on external timing using a slave computer and a photocell (e.g., De Clercq, Crombez, Buysse, & Roeyers, 2003). Compatibility of Affect 4.0 with the VISTA OS is currently under investigation.
Affect 4.0 can be downloaded from
To use Affect 4.0, simply store the desired .exe file on your hard drive in a folder that also contains the inpout.dll (freeware for noncommercial use). This .dll is also included in the Affect 4.0 download.
Disclaimer, Bug Reports, and Feature Requests
>
Although Affect 4.0 has been carefully tested, the authors disclaim any responsibility for possible errors. Nevertheless, bugs can be reported at the following website:
Footnotes
<anchor name="fn1"></anchor><sups> 1 </sups> In principle, timing accuracy may be compromised if one would specify an abundant number of trial commands to be executed at the same flow event. This problem will rarely occur in practice, however, as the number of trial commands that are needed to implement any particular trial flow is typically low. We will return to the issue of timing accuracy below.
<anchor name="fn2"></anchor>
<sups>
2
</sups> Whereas trial properties are usually defined in terms of stimulus properties (see above), one can also define trial properties as a function of preceding semirandomizations. That is, experimenters can first semirandomly select a number of stimuli or stimulus combinations and then specify whether this set of stimuli or stimulus combinations exhibits certain trial properties on a post hoc basis. For example, if one were to design an affective priming experiment in which prime and target stimuli are presented simultaneously at
References
<anchor name="c1"></anchor>Bradley, B., Field, M., Mogg, K., & De Houwer, J. (2004). Attentional and evaluative biases for smoking cues in nicotine dependence: Component processes of biases in visual orienting.
Coppens, E., Vansteenwegen, D., Spruyt, A., Vandenbulcke, M., Van Paesschen, W., & Eelen, P. (2007). Automatic affective stimulus processing is intact after unilateral resection of the temporal lobe.
Crosbie, J. (1990). The Microsoft mouse as a multipurpose response device for the IBMPC/XT/AT.
Curtin, J. J., Lozano, D. L., & Allen, J. J. B. (2007). The psychophysiology laboratory. In J. A.Coan & J. J. B.Allen (Eds.),
De Clercq, A., Crombez, G., Buysse, A., & Roeyers, H. (2003). A simple and sensitive method to measure timing accuracy.
De Clercq, A., Verschuere, B., De Vlieger, P., & Crombez, G. (2006). Psychophysiological Analysis (PSPHA): A modular script-based program for analyzing psychophysiological data.
De Cort, K., Hermans, D., Spruyt, A., Griez, E., & Schruers, K. (2008). A specific attentional bias in panic disorder?.
Fazio, R. H., Sanbonmatsu, D., Powell, M., & Kardes, F. (1986). On the automatic activation of attitudes.
Gawronski, B., Deutsch, R., & Seidel, O. (2005). Contextual influences on implicit evaluation: A test of additive versus contrastive effects of evaluative context stimuli in affective priming.
Klauer, K. C., & Musch, J. (2002). Goal-dependent and goal-independent effects of irrelevant evaluations.
Klinger, M. R., Burton, P. C., & Pitts, G. S. (2000). Mechanisms of unconscious priming: I. Response competition, not spreading activation.
Malley, G. B., & Strayer, D. L. (1995). Effect of stimulus repetition on positive and negative identity priming.
Mogg, K., Bradley, B. P., Field, M., & De Houwer, J. (2003). Eye movements to smoking related pictures in smokers: Relationship between attentional biases and implicit and explicit measures of stimulus valence.
Musch, J., & Klauer, K. C. (2001). Locational uncertainty moderates affective congruency effects in the evaluative decision task.
Myors, B. (1999). Timing accuracy of PC programs running under DOS and Windows.
Neely, J. H. (1991). Semantic priming effects in visual word recognition: A selective review of current findings and theories. In D.Besner & G. W.Humphreys (Eds.),
Öhman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention: Detecting the snake in the grass.
Plant, R. R., Hammond, M., & Turner, G. (2004). Self-validating presentation and response timing in cognitive paradigms: How and why?.
Pratto, F., & John, O. P. (1991). Automatic vigilance: The attention-grabbing power of negative social information.
Segalowitz, S. J., & Graves, R. E. (1990). Suitability of the IBM XT, AT, and PS/2 keyboard, mouse, and game port as response devices in reaction time paradigms.
Spruyt, A., De Houwer, J., Hermans, D., & Eelen, P. (2007). Affective priming of non-affective semantic categorization responses.
Spruyt, A., Hermans, D., De Houwer, J., & Eelen, P. (2004). Non-associative semantic priming: Episodic affective priming of naming responses.
Spruyt, A., Hermans, D., De Houwer, J., Vandekerckhove, J., & Eelen, P. (2007). On the predictive validity of indirect attitude measures: Prediction of consumer choice behavior on the basis of affective priming in the picture-picture naming task.
Spruyt, A., Hermans, D., De Houwer, J., Vandromme, H., & Eelen, P. (2007). On the nature of the affective priming effect: Effects of stimulus onset asynchrony and relatedness proportion in naming and evaluative categorization.
Stahl, C. (2006). Software for generating psychological experiments.
Strayer, D. L., & Grison, S. (1999). Negative identity priming is contingent on stimulus repetition.
Tipper, S. P. (1985). The negative priming effect: Inhibitory priming by ignored objects.
Vansteenwegen, D., Iberico, C., Vervliet, B., Marescau, V., & Hermans, D. (2008). Contextual fear induced by unpredictability in a human fear conditioning preparation is related to the chronic expectation of a threatening US.
Vansteenwegen, D., Vervliet, B., Iberico, C., Baeyens, F., Van den Bergh, O., & Hermans, D. (2007). The repeated confrontation with videotapes of spiders in multiple contexts attenuates renewal of fear in spider anxious students.
Voss, A., Leonhart, R., & Stahl, C. (2007). How to make your own response boxes: A step-by-step guide for the construction of reliable and inexpensive parallel-port response pads from computer mice.
Wan, L., Van Diest, I., De Peuter, S., Bogaerts, K., Oyen, N., & Hombroux, N., et al. (2008). Repeated experiences of subjective air hunger and ventilatory behavior in response to hypercapnia in the Standardized Rebreathing Test: Effects of anxiety.
Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology.