Posted: February 26th, 2023

PSYC 575 Cognitive Psychology

 Journal Article Summary: Problem-Solving and Expertise 

  • Priming analogical reasoning with false memories
  • Mark L. Howe & Sarah R. Garner & Emma Threadgold &

    Linden J. Ball

    Published online: 18 March 2015
    # The Author(s) 2015. This article is published with open access at Springerlink.com

    Abstract Like true memories, false memories are capable of
    priming answers to insight-based problems. Recent research
    has attempted to extend this paradigm to more advanced
    problem-solving tasks, including those involving verbal ana-
    logical reasoning. However, these experiments ar

    e

    constrained inasmuch as problem solutions could be generat-
    ed via spreading activation mechanisms (much like false
    memories themselves) rather than using complex reasoning
    processes. In three experiments we examined false memory
    priming of complex analogical reasoning tasks in the absence
    of simple semantic associations. In Experiment 1, we demon-
    strated the robustness of false memory priming in analogical
    reasoning when backward associative strength among the
    problem terms was eliminated. In Experiments 2a and 2b,
    we extended these findings by demonstrating priming on new-
    ly created homonym analogies that can only be solved by
    inhibiting semantic associations within the analogy. Overall,
    the findings of the present experiments provide evidence that
    the efficacy of false memory priming extends to complex an-
    alogical reasoning problems.

    Keywords Priming . Analogical reasoning . False memory .

    DRMparadigm . Homonyms . Spreading activation

    Memory is highly flexible and reconstructive, designed to
    retain information about the past, interpret the present, and

    support simulations of future events (e.g., Howe, 2011;
    Newman & Lindsay, 2009; Schacter, Guerin, & St. Jacques,
    2011). Interestingly, recent research has shown that memory is
    highly functional, regardless of whether we are talking about
    memories for events that actually occurred (i.e., true memo-
    ries) or for self-generated memories of events that did not
    occur (i.e., false memories). For example, a significant body
    of research has demonstrated that true memories are able to
    prime performance on related memory tasks (e.g., implicit
    memory; see Gulan & Valerjev, 2010) as well as non-
    memory tasks such as verbal problem solving (e.g.,
    Mednick, Mednick, & Mednick, 1964).

    Priming refers to Ba change in the ability to identify, pro-
    duce, or classify an item as a result of a prior encounter with
    that item, or a related item^ (Schacter, Gallo, & Kensinger,
    2007, p. 356). In the case of analogical reasoning, for exam-
    ple, there is a well-established body of evidence demonstrat-
    ing that people are able to transfer directly their prior memo-
    ries of problems and their solutions in order to assist them in
    tackling new, related problems (e.g., Bassok & Holyoak,
    1989; Richland, Zur, & Holyoak, 2007; for a recent review,
    see Holyoak, 2012). Although such analogical reasoning pro-
    cesses appear to rely largely on direct or explicit memory
    retrieval, there is also evidence that prior memories can influ-
    ence reasoning and problem solving through intuitive mecha-
    nisms that operate indirectly or implicitly. Such intuitive pro-
    cesses appear to have a basis either in tacitly learned memory
    associations (e.g., Osman & Stavy, 2006; Sloman, 1996) or in
    rules that were once deliberatively acquired but which have
    been practiced so extensively that they have reached a state of
    automaticity in procedural memory (e.g., Kahneman & Klein,
    2009).

    Kokinov (1990; Kokinov & Petrov, 2001), for example,
    has shown that implicit memory priming can facilitate perfor-
    mance with complex deductive, inductive, and analogical rea-
    soning problems, benefitting both the strategy taken and the

    M. L. Howe (*) : S. R. Garner : E. Threadgold
    Department of Psychology, City University London, Northampton
    Square, London EC1V 0HB, UK
    e-mail: Mark.Howe.1@city.ac.uk

    L. J. Ball
    School of Psychology, University of Central Lancashire, Preston PR

    1

    2HE, UK

    Mem Cogn (2015) 43:879–895
    DOI 10.3758/s13421-015-0513-7

    success of the problem-solving process. Schunn and Dunbar
    (1996) have provided further support for priming effects in
    analogical problem solving, demonstrating that conceptual
    knowledge of one knowledge domain (i.e., biochemistry)
    can spontaneously influence complex reasoning in another,
    unrelated knowledge domain (i.e., molecular genetics) via im-
    plicit priming, leading to facilitated problem solving as mea-
    sured through both accuracy and the speed of solution gener-
    ation. Schunn and Dunbar’s sophisticated controls and mea-
    sures also allowed for any involvement of explicit memory
    processes to be ruled out as a cause of solution success in the
    implicit priming conditions.

    More recently, it has been discovered that it is not just true
    memories that can prime performance on cognitive tasks such
    as problem solving but that false memories can also have key
    beneficial effects. A common procedure used to generate false
    memories is the Deese/Roediger-McDermott (DRM) para-
    digm (Deese, 1959; Roediger & McDermott, 1995). Here,
    participants are presented with words (e.g., snooze, doze,
    wake, rest) that are all semantic associates of a non-
    presented word or so-called critical lure (e.g., sleep). When
    asked to remember the words on the list, participants frequent-
    ly remember the critical lure (a false memory) along with the
    presented items. Using this paradigm, it has been shown that
    false memories can prime solutions to problem-solving tasks
    such as insight-based Compound Remote Associate Tasks
    (CRATs; see Howe, Garner, Dewhurst, & Ball, 2010) and
    verbal proportional analogies (Howe, Threadgold, Norbury,
    Garner, & Ball, 2013).

    The latter problems (i.e., verbal proportional analogie

    s)

    involve the presentation of items that have the form a is to b
    as c is to d, where participants are given the a, b, and c terms
    and are asked to generate the missing d term (e.g., Ball, Hoyle,
    & Towse, 2010; Goswami, 2001; Goswami & Brown, 1989,
    1990). For example, the participant might be given the prob-
    lem ‘dog is to kennel as bird is to ?’ and asked to generate the
    solution term. The optimal way to solve such analogies in-
    volves identifying the relation that exists between the a and
    b terms (in this case, ‘inhabits’) and thenmapping this relation
    onto the c term (‘bird’) in order to generate the answer ‘nest.’
    Proportional analogies of this type are non-trivial, especially
    for children, but even adult performance is rarely error free
    (e.g., see Green, Fugelsang, & Dunbar, 2006). Such problems
    therefore frequently feature in intelligence tests (Sternberg,
    1977) and in academic examinations such as the statutory
    assessment test.

    Although non-trivial, proportional analogies are typically
    easier for adults to solve than are many other forms of com-
    plex analogy problems that have been studied in the literature,
    which tend to involve the identification and mapping of mul-
    tiple, hierarchically embedded ‘systems’ of relations (for
    pioneering research with such problems, see Gentner &
    Toupin, 1986; Gick & Holyoak, 1983; Keane, 1987). We

    acknowledge that proportional analogies do not involve the
    sophistication of complex analogies of the type that have dom-
    inated much of the analogical reasoning literature, and that are
    typically very challenging for adults to solve in the absence of
    directive hints to use specific past experiences. Nevertheless, a
    major advantage of studying false memory priming effects
    with proportional analogies derives from the way in which
    such problems afford an opportunity to impose very strict
    controls on the terms that they are composed of. As will be
    shown in the experiments that we present below, such controls
    facilitate the examination of some unique aspects of analogi-
    cal problem solving that have hitherto remained unexplored.

    Although it has been established that both true and false
    memories can effectively prime solutions to problem-solving
    tasks (including proportional analogies), an interesting devel-
    opment has been that false memories can actually be more
    effective primes for problem solving than true memories
    (Howe et al., 2013; Howe, Wilkinson, & Monaghan, 2012;
    Wilkinson, 2014). This is consistent with the literature
    documenting the different strengths of true and false memo-
    ries where the latter have been shown to be stronger than the
    former (e.g., Brainerd, Reyna, & Brandse, 1995; Howe,
    Candel, Otgaar, Malone, & Wimmer, 2010a; McDermott,
    1996). For example, whereas truememories decline over time,
    false memories persist across retention intervals (days, weeks;
    Brainerd et al., 1995; McDermott, 1996) and negative false
    memories can actually increase over time (e.g., Howe et al.,
    2010a).

    That false memories can be stronger than true memories
    has been attributed to the different ways in which they are
    formed. Specifically, false memories tend to be self-generated
    (i.e., occurring spontaneously and automatically as a result of
    internal semantic activation) whereas true memories are often
    other-generated (e.g., presented on a list by the experimenter).
    This self- versus other-generated difference holds regardless
    of the nature of the paradigm being used and has been ob-
    served using the standard DRM paradigm (e.g., Howe, 2005),
    when participants are remembering stories, pictures, and
    videos (e.g., Otgaar, Howe, Peters, Sauerland, &
    Raymaekers, 2013; Otgaar, Howe, Peters, Smeets, & Moritz,
    2014), and when entire memories are being implanted (e.g.,
    Otgaar, Smeets, & Peters, 2012). The efficacy of self-
    generated information is underpinned by a substantial body
    of research showing that self-generated information is not on-
    ly encoded at a deeper level but is also significantly more
    likely to be remembered than other-generated (i.e., experi-
    menter presented) information (Bjorklund, 2004; Slamecka
    & Graf, 1978). Thus, if priming effects are monotonically
    related to memory strength, then false memories should be
    better primes than true memories, particularly following a de-
    lay. The benefit of falsely remembered items in priming solu-
    tions to problem-solving tasks (e.g., CRATs) has been
    established both with adults (Howe et al., 2010b) and children

    880 Mem Cogn (2015) 43:879–895

    (Howe, Garner, Charlesworth, & Knott, 2011). Howe et al.
    (2013) attempted to extend this effect to more complex rea-
    soning tasks by using false memories to prime solutions to
    analogical reasoning problems. Like the research with
    CRATs, both adults and children were primed on verbal pro-
    portional analogies of the form a is to b as c is to d and were
    asked to generate the d term. The solution to six of the nine
    verbal analogies was also the critical lure from previously
    presented DRM lists (e.g., desert is to hot as arctic is to cold,
    where cold was both the solution to the analogy and the crit-
    ical lure of a DRM list). For the six analogies that were primed
    (the remaining three were not primed), three were primed by
    having the critical lure presented as a list item (a ‘true’ or
    other-generated memory) and the remaining three were
    primed by not having the critical lure as a list item (a ‘false’
    or self-generated memory). The results showed that, unsur-
    prisingly, adults solved the analogies more quickly than chil-
    dren. Importantly, both adults and children solved verbal anal-
    ogies more quickly when primed with a false memory than
    when unprimed or when primed by a true memory (there were
    no differences between the latter two conditions).

    Although these effects for false memory priming of analo-
    gies are interesting, they are also somewhat limited. This is
    because Howe et al. (2013) used relatively straightforward
    analogies that were solved quickly and easily by children
    and adults alike. Although this allowed for a demonstration
    of priming effects in both adults and children, a downside is
    that priming in this context represents activation of simple
    semantic associates and not the priming of complex reasoning
    relations themselves.

    To explain, the distinction between the priming of simple
    semantic associations versus the priming of more complex,
    analytic problem solving is of particular concern in the verbal
    analogies literature, where a debate exists concerning the
    mechanisms by which proportional analogies are solved.
    Some researchers (e.g., Green et al., 2006) suggest that pro-
    portional analogies are solved analytically in the optimal man-
    ner described above, which involves mapping the relation
    between the a and b terms onto the c term in order to generate
    the answer d. Others (e.g., Sternberg & Nigro, 1980), howev-
    er, have proposed that proportional analogies are typically
    solved using semantic associations (particularly by children;
    e.g., see Ball et al., 2010; Cheshire, Muldoon, Francis, Lewis,
    & Ball, 2007; Siegler & Svetina, 2002) in a similar manner to
    the spreading activation processes thought to underlie the so-
    lutions to CRATs. Given the relatively simplistic nature of
    Howe et al.’s (2013) verbal analogies, solutions could have
    been generated using associations generated via spreading ac-
    tivation. These analogies could be solved analytically through
    relational mapping, but given the high semantic association
    between the c and d terms, it was equally possible that these
    analogies provided more of a semantic or word association
    task than a true test of analogical reasoning. This could mean

    that what Howe et al. (2013) demonstrated was not the ability
    of false memories to prime analogical reasoning via a
    relational-mapping process but simply their ability to prime
    closely related semantic associations (e.g., where the b term
    ‘hot’ or the c term ‘arctic’ simply primed the d term ‘cold’).
    Therefore, a task is needed that can be used to demonstrate the
    ability of false memories to prime the solutions to complex
    reasoning problems in the absence of simple semantic
    associations.

    The purpose of the present research was therefore twofold.
    First, we wanted to develop new analogical reasoning tasks,
    ones that rely less heavily on simple semantic associations and
    instead are more dependent on analytic, relational mapping.
    Second, we wanted to investigate whether false memories are
    still capable of priming the solutions to these complex analog-
    ical reasoning tasks when these solutions rely less heavily on
    spreading activation among a single set of semantic associa-
    tions. In order to do this, we have developed two new sets of
    analogical reasoning tasks.

    In Experiment 1, we created a set of verbal proportional
    analogies that are considerably less semantically related than
    those used in the previous experiment (Howe et al., 2013).
    Specifically, by controlling backward associative strength
    (BAS; a numerical measure of the likelihood that a target word
    will be produced given a cue word) it was possible to reduce
    (or in most cases eliminate) the semantic relationship between
    the a to d, b to d, and c to d terms. We then calculated an
    overall cumulative BAS score for the target word d (solution)
    being produced as a simple associate of the cue words (a, b,
    and c terms) provided in the analogy. The lower this cumula-
    tive BAS score, the less likely the analogical problem is to be
    solved by spreading activation of associations in memory
    from the analogy terms alone, independent of analytic reason-
    ing. When we calculated cumulative BAS for the analogies
    used in Howe et al. (2013), the value was 3.94. In contrast, the
    cumulative BAS of the analogies used in our first experiment
    was 0.23. A one-way analysis of variance (ANOVA) con-
    firmed that there was a significant difference between the cu-
    mulative BAS of the analogies in the experiment reported by
    Howe et al. and those used in our Experiment 1. Thus,
    Experiment 1 provides a more appropriate demonstration of
    false memory priming of analogy problems requiring true an-
    alogical reasoning rather than problems that merely tap into
    semantic associations of memory, such as those reported in
    Howe et al. (2013).

    In Experiments 2a and 2b, we extended the priming of
    analogical reasoning based around an analytic mapping pro-
    cess (as opposed to simple semantic associations) by develop-
    ing a new type of analogical reasoning task called a homonym
    analogy task. In this task, we used homonyms, which are
    words that are pronounced the same but have very different
    contextual meanings (e.g., words such as score). In this way
    we could ensure that analogies were more likely to be solved

    Mem Cogn (2015) 43:879–895 881

    using analytic mapping of the relational term and not just
    spreading activation among semantic associations.

    Experiment 1

    In Experiment 1, we investigated false memory priming of
    verbal proportional analogies using a set of normed analogical
    reasoning problems in which we limited the cumulative BAS
    of the terms provided in the analogical problem.

    Method

    Participants

    The participants were twenty-five 18-year-old undergraduate
    students who were fluent in English. Recruitment took place
    via a participant recruitment system, and each participant re-
    ceived £3.50 for 30 minutes of participation time. Written
    informed consent was obtained from each participant prior
    to taking part in the experiment, and participants were
    debriefed following their participation.

    Design and materials

    A within-participant design was employed consisting of one
    factor with two levels (Priming: Unprimed or

    False

    Memory

    Priming). The experiment was programmed using Psyscript
    (an experimental generator) and run by an Apple Macintosh
    computer. Eight normed proportional analogical reasoning
    problems (of the format a is to b as c is to d) were used in this
    experiment (see Table 1). These analogies were a subset se-
    lected from a previous norming study (Howe, Threadgold,
    Garner, Bland, & Ball, 2015) in which we asked 50 partici-
    pants to generate the answers to 50 newly created proportional
    analogies, with a maximum of 60 seconds being given for the
    generation of an answer to each problem, after which the

    correct solution was displayed. Analogies were selected for
    the present experiment if their normed solution rate fell be-
    tween 20 % and 80 %, and if the strength of the BAS of the
    associated DRM list allowed for the attainment of effective
    experimental controls, as discussed below.

    The subset of eight analogical reasoning problems that
    were selected (see Table 1) had normed solution rates in the
    range of 34 % to 76 %. These eight problems were divided
    into two groups of four analogies, with the presentation of
    these groups being counterbalanced across participants in the
    experiment in terms of whether they were unprimed or primed
    by the prior presentation of a DRM list. The four analogies in
    each group were equated on the BAS of the DRM list items
    and on their normed solution rates (Group 1 analogies—earth,
    lion, stone, iron: Mean DRM BAS = .20, Mean Solution Rate
    = 61.5 %; Group 2 analogies – allow, spider, needle, hair:
    Mean DRM BAS = .14, Mean Solution Rate = 53.5 %).
    Because BAS is a widely used measure of the strength of a
    DRM list in producing false memories, it is important to con-
    trol for Mean DRM BAS level across conditions (e.g.
    Roediger, Watson, McDermott, & Gallo, 2001).

    Furthermore, to provide an indication of the semantic
    strength of the analogy the BAS of the a to d, b to d, and c
    to d relationships (e.g., the likelihood of producing the solu-
    tion d when asked to provide a semantic associate of a, b, and
    c) were totaled, providing a cumulative BAS score for each
    analogical problem. All BAS values were selected from the
    normed associates presented by Nelson, McEvoy, and
    Schreiber (1998). There was no significant difference
    in the cumulative BAS for the a to d, b to d, and c
    to d relationships between each group of four analogies
    (Cumulative BAS: Group 1 = .06, Group 2 = .17, p >
    .05). All eight analogies had a zero BAS score for the a
    to d and b to d relationships. Three analogies (those
    with the solutions iron, hair, and needle) had an above
    zero, but still very low, c to d BAS score (.06, .14, and
    .03, respectively). Overall, then, there was a very low
    likelihood of the a and b cue words in any of the
    analogies producing solution words by spreading activa-
    tion alone.

    For each analogical reasoning problem there was a linked
    DRM list consisting of 12 associated words where the critical
    lure was also the solution to the problem (see Appendix A for
    DRM lists and BAS scores for each list). DRM lists contain-
    ing 12 associate terms were used. The use of 12 associates is
    consistent with early applications of the DRM paradigm
    (Deese, 1959) and has been frequently shown to induce false
    recall of the critical lure (Roediger & McDermott, 1995). The
    DRM lists were either selected from standard sources (e.g.,
    Roediger et al., 2001) or were constructed based on the
    normed associates presented by Nelson et al. (1998). Words
    on the DRM lists did not appear as part of the analogical
    reasoning problems.

    Table 1 Mean solution rates and times (with standard deviations in
    parenthesis) for the normed proportional analogies used in Experiment 1

    Analogy with solution Mean Solution
    rate

    Mean solution
    time (s)

    peace : dove :: courage : lion 0.76 (.43) 4.56 (3.35)

    prevent : restrict :: enable : allow 0.74 (.44) 6.02 (3.71)

    car : roundabout :: moon : earth 0.62 (.49) 5.99 (3.25)

    four : cat :: eight : spider 0.60 (.49) 5.45 (4.10)

    egg : yolk :: plum : stone 0.56 (.50) 4.57 (2.32)

    wash : clean :: press : iron 0.52 (.51) 7.08 (4.62)

    leopard : spots :: chest : hair 0.46 (.50) 9.31 (6.45)

    watch : cog :: compass : needle 0.34 .(47) 5.11 (2.43)

    882 Mem Cogn (2015) 43:879–895

    Procedure

    Participants were informed that they would be completing two
    distinct tasks: a memory task and problem-solving task.
    Therefore, it was never explicitly stated to participants that
    the word lists were linked to the analogical reasoning prob-
    lems in any way, or that the memory task could be used to help
    solve the analogical reasoning problems.1 Participants initially
    listened to four DRM lists played to them through headphones
    via a computer. DRM items were presented at the rate of one
    word every two seconds. Lists were played individually and in
    a random order for each participant, but the order of the items
    in the list remained constant for each participant. Following
    presentation of a list there was a brief filler task (consisting of
    two simple arithmetic calculations on screen) before partici-
    pants were asked to write down as many words as they could
    remember on a piece of paper, provided. Participants were not
    given a time limit to recall these words and were merely
    instructed to proceed when they had recalled as many words
    as they could from the given list. Following the completion of
    all four DRM lists, participants were asked to turn their paper
    over so they could not see their recall answers before
    attempting to solve eight analogical problems, presented one
    at a time to them on the computer screen.

    Presentation of the eight analogical problems was random-
    ized for each participant. Analogical problems were presented
    in the format ‘a is to b as c is to _____’ in the center of the
    computer screen. Participants were required to click a button
    as soon as they had their final answer to an analogy, and they
    then needed to type their answer into the space provided on
    screen. The timer began as soon as participants viewed the
    analogy on screen and ended once participants had clicked
    the button signaling that they had their final answer.
    Participants received a maximum of 60 seconds to generate
    each answer, after which the correct answer was displayed. On
    providing their answer to each analogy, participants viewed
    the complete analogy with the correct answer on screen.
    Participants completed eight analogies in total, four of which
    had been primed by the associated DRM list and four of which
    were unprimed.

    Results

    The mean analogy solution rate (proportion) and the mean
    analogy solution time (seconds) were calculated for each par-
    ticipant and analyzed in separate ANOVAs. For the primed
    analogical reasoning problems, solution rates and times were
    further conditionalized according to whether a false memory
    had been produced during recall of the DRM list relevant to

    that analogy. Conditionalizing primed performance in this
    manner has been widely used in previous research investigat-
    ing the priming capacity of false memories (e.g., Howe et al.,
    2010b, 2013). Despite the reduction in items per cell when
    responses are conditionalized (although there was still suffi-
    cient power to detect differences should they exist as the ma-
    jority of participants, over 65 %, contributed data to all three
    cells), it is imperative that this distinction is made because
    previous research has consistently shown that priming is only
    effective when the false memory is actually produced at on a
    memory test. Therefore, there were three levels of priming for
    the analyses that we report below: unprimed vs. false memory
    primed with no false recall vs. false memory primed with false
    recall. The mean false memory proportion was .26 (SD = .13)
    with the majority of participants (84 %) having one or more
    false memories.

    Solution rates

    There was a significant main effect of priming for solution
    rates, F(2, 24) = 6.17, p < .05, η2p = .34. As can be seen in
    Fig. 1, and which was confirmed using post hoc pairwise
    comparisons, solution rates were significantly higher in the
    false memory priming condition when a false memory had
    been produced at recall (M = .94, SE = .04) compared to either
    false memory priming where no false memory was produced
    (M = .60, SE = .02, p < .05) or the unprimed condition (M =
    .62, SE = .06, p .05).

    Solution times

    Like the solution rate data, there was a significant main effect
    of priming for solution times, F(2, 20) = 4.72, p < .05, η2p =
    .32. As can be seen in Fig. 2, and which was confirmed using
    post hoc pairwise comparisons, solution times were signifi-
    cantly faster in the false memory priming condition when a
    false memory had been produced (M = 3.79 s, SE = .43)

    1 In fact, when participants were later asked during debriefing whether
    they thought the two tasks were in any way related, over 95 % say that
    they did not believe that the two tasks were connected.

    0

    0.

    2

    0.

    4

    0.

    6

    0.

    8

    1

    Unprimed Primed No

    False Memory

    Primed False

    Memory

    M
    ea

    n
    P

    ro
    p

    o
    rt

    i

    o
    n

    S
    o

    lu
    ti

    o
    n

    R
    at

    e

    Priming Condition

    Fig. 1 Mean proportion of solution rates (with standard errors) as a
    function of priming condition

    Mem Cogn (2015) 43:879–895 883

    compared to either the false memory priming with no false
    memory (M = 9.05 s, SE = 1.51, p < .05) or the unprimed
    conditions (M = 10.54 s, SE = 2.77, p .05).

    Discussion

    The results of Experiment 1 are unique inasmuch as they show
    that having a false memory is critical for the priming of ana-
    logical reasoning that requires a relational mapping process to
    arrive at a solution. That is, only when participants recalled the
    critical lure did false memories prime the solutions to verbal
    proportional analogies. When participants failed to recall the
    critical lure, performance (solution rates and solution times)
    on analogical reasoning problems was no better than when
    solutions had not been primed. Thus, we can conclude that
    the production of the critical lure is imperative for the success
    of a false memory priming effect in analogical reasoning.

    These findings also provide an important and unique dem-
    onstration of the benefit of false memories, one that extends
    our knowledge of their ability to prime performance not just
    on related memory tasks but on more complex problem-
    solving tasks as well. Although previous research has been
    influential in establishing evidence for the ability of false
    memories to prime the solutions to insight problems (e.g.,
    Howe et al., 2010b, 2011), this is the first experiment to ex-
    tend these findings to more complex analogical reasoning
    tasks, ones that require a process of analogical mapping, and
    that cannot be solved solely by activating simple spreading
    activation among semantic associations (as was the case in
    Howe et al., 2013).

    Making this distinction is particularly important for theo-
    ries of analogical reasoning, where a debate exists concerning
    the mechanisms by which proportional analogies are solved.
    Here, some researchers argue that analogies are solved by a
    process of semantic association and not by using analogical

    mapping. The results of the present experiment suggest that
    when one limits the availability of semantic associations be-
    tween the analogy terms and the solution, it is still possible
    both to solve these analogies (60 % of the time) and, impor-
    tantly, to prime these solutions using false memories.

    Experiment 2a

    Experiment 1 utilized a set of normed verbal proportional
    analogies in which we limited the likelihood of the target
    solution being arrived at using spreading activation through
    semantic associates. However, despite this control, one could
    argue that this association, rather than being removed, was
    simply more remote than in the analogies used previously by
    Howe at al. (2013). That is, the analogies presented in
    Experiment 1 might still have been solved via spreading acti-
    vation, albeit requiring the activation of more distant or weak-
    er associations (for how such a process might work, see
    Nelson, Kitto, Galea, McEvoy, & Bruza, 2013; Nelson,
    McEvoy, & Pointer, 2003).

    If semantic association did still play a role in solving ana-
    logical reasoning problems in Experiment 1, what would hap-
    pen if we were to remove the influence of associations by
    ensuring that the proportional analogy could only be solved
    via an analytic mapping process? Moreover, what would hap-
    pen if analogical problem solving purposefully required par-
    ticipants to inhibit any semantic associations that may be ap-
    parent within the proportional analogy?

    In order to examine these questions, we developed a task
    that not only removed the use of semantic association as a
    solution strategy but also required the inhibition of dominant
    semantic associations within the analogy problem in order
    obtain the solution using an analytic mapping process. We
    did this by creating a new type of verbal proportional analogy
    termed a homonym analogy. Homonym analogies take the
    standard form of a verbal proportional analogy, that is, a is
    to b as c is to d. For example, ‘fur is to bear as bark is to tree.’
    However, these analogies differ in a number of important
    ways to the standard verbal proportional analogies used in
    Experiment 1.

    To see how these homonym analogies work, consider the
    following example: fur is to bear, as bark is to: __? Rather
    than have participants solve these analogies in the usual man-
    ner, we gave participants four multiple-choice options, in this
    case, branch, dog, meow, and tree (see Table 2). In this exam-
    ple, the a and b terms, fur and bear, create a context related to
    the category of animals, making one likely to use this context
    when interpreting the ambiguous homonym c term, bark. If
    participants are biased by this context, they will interpret the c-
    term bark in terms of an animal (the noise a dog makes) and
    select the answer dog. In the multiple-choice options, we in-
    cluded two incorrect but contextually relevant associates (one

    0

    2

    4

    6

    8

    10

    12

    14

    Unprimed Primed No False

    Memory

    Primed False

    Memory

    M
    ea

    n
    S

    o
    lu

    ti
    o

    n

    T
    im

    es
    (

    s)

    Priming Condition

    Fig. 2 Mean solution times (seconds) with standard errors as a function
    of priming condition

    884 Mem Cogn (2015) 43:879–895

    high associate and one low associate: dog and meow, respec-
    tively) to determine whether participants are biased towards
    using semantic associations rather than relational mapping to
    solve these problems. Alternatively, when given the multiple-
    choice options, if one reasons correctly that fur is the outside
    of a bear, one will apply this relation analytically to the c term,
    the homonym bark, and correctly reason the solution that bark
    is the outside of a tree. A further associate of this correct
    context is also provided, other than the correct answer, which
    in this example is the word branch. If participants incorrectly
    select this solution word during the task it would suggest that
    although they are able to inhibit the incorrect meaning of the
    homonym (or even interpret the homonym with the required
    meaning to solve the problem, without any consideration of
    the a and b terms) they might still reason incorrectly.

    In summary, there are a number of critical differences be-
    tween standard proportional analogies and our newly created
    homonym analogies. The first important difference is that the
    c term used in these new analogies is a homonym, that is, a
    term that can have multiple meanings in different contexts and
    which is therefore ambiguous in nature. Second, the a and b
    terms in a given homonym analogy set a context related to one
    of the meanings of the c term, specifically, a context that is not
    related to the solution to the analogy. Third, the d term used to
    solve the analogy requires participants to inhibit the context
    created by the a and b terms and to access the alternative
    meaning of the c term in order to achieve the solution.

    A final difference is that participants are asked to select the
    analogy solution from among four multiple-choice options.
    This is a deviation from the methodology employed in
    Experiment 1, in which participants were asked to generate
    the d term response to standard verbal proportional analogies.
    A multiple-choice response paradigm was adopted with the
    homonym analogies so that it was possible to analyze specific
    types of errors provided to these problems by participants. The
    options consisted of the correct solution and three incorrect
    choices that were carefully selected to fall into one of three
    categories: (1) a correct context associate – a term that is
    semantically related to the correct solution, which also re-
    quires participants to access the correct meaning of the hom-
    onym; (2) an incorrect context high associate – a term that is
    highly semantically associated to the homonymwhen taken in
    the context of the a and b terms of the analogy, but which is

    incorrect when one achieves an effective relational mapping
    from the a and b terms to the c term; and (3) an incorrect
    context low associate – a term that is a low semantic associate
    of the homonym when taken in the context of the a and b
    terms of the analogy, but which is again incorrect when one
    achieves an effective relational mapping from the a and b
    terms to the c term.

    When constructing the multiple choice items from which
    participants selected their final answer, written word frequen-
    cy was controlled using the Kucera-Francis written word fre-
    quency scores obtained from the MRC Psycholinguistic
    Database (Coltheart, 1981). A highest word frequency item
    from the four that were presented occurred as a critical item
    and the incorrect context high associate on three instances, and
    as a correct context associate and incorrect context low asso-
    ciate on two instances each. Therefore, any multiple choice
    answer type (see Table 2 for examples) was not likely to be
    consistently selected based on dominant written-word fre-
    quency alone.

    Given the multiple contextual interpretations of homonym
    terms, it is important to consider the dominance of any single
    homonym context in comparison to its counterpart meanings.
    To do this, we consulted Twilley, Dixon, Taylor, and Clark’s
    (1994) frequency norms for the different meanings of our
    critical homonym c terms. Drawing on the previously utilized
    analogy example with the homonym bark, Twilley et al.
    (1994) noted the primary context is in terms of dog and sec-
    ondarily in terms of tree. Therefore, the overall analogy is
    consistent with the dominant homonym context (animals or
    dog), while the correct answer context (bark in terms of the
    outer lining of a tree) is consistent with the second dominant
    meaning of the homonym. We consider the effects of hom-
    onym dominance in terms of responses to homonym analogies
    with Experiment 2a. In Experiment 2a, we collected norms for
    these new types of problems by having participants solve a set
    of homonym analogies.

    Method

    Participants

    Fifty-six participants (9males; 47 females) aged 16 to 18 years
    old took part in this experiment. All participants aged 18 and
    over provided written informed consent prior to the experi-
    ment, and for those aged 16 or 17 years, parental consent was
    sought prior to participation. At the end of the experiment, all
    participants were fully debriefed about the purpose of the
    experiment.

    Design

    A within-participant design was used, with each participant
    completing a selection of three out of nine homonym

    Table 2 Multiple choice answers for the homonym analogy ‘fur is to
    bear, as bark is to tree.’

    Multiple choice answers Example item

    Correct answer Tree

    Correct context associate Branch

    Incorrect context high associate Dog

    Incorrect context low associate Meow

    Mem Cogn (2015) 43:879–895 885

    analogies. The order of analogies was randomized. Because
    the answer to the homonym analogy is the opposite context to
    that presented in the analogy, this may have become apparent
    to participants after solving a number of them. Participants
    were therefore asked to complete three out of nine analogies
    in order to prevent practice effects at solving these problems.
    This issue was addressed in Experiment 2b by the introduction
    of non-homonym analogies similar to those used in
    Experiment 1, which served as distractor problems.

    Materials and procedure

    The analogical reasoning problems utilized in this experiment
    were the newly formed homonym analogies. Nine such anal-
    ogies were created (see Appendix B). For each analogical
    reasoning problem participants were provided with four pos-
    sible answers to choose from. One of these was the correct
    solution, and then there were also three possible foils: a correct
    context associate, an incorrect context high associate, and an
    incorrect context low associate.

    The procedure was identical to the problem-solving com-
    ponent in Experiment 1, with the exception that participants
    completed three of nine homonym analogies. Furthermore,
    participants were asked to choose an answer from one of four
    provided. These options were displayed directly underneath
    the analogy and labeled a to d. The position of the correct
    answer in terms of a to d was randomized.

    Results

    The percentage of participants solving each homonym analo-
    gy within the time limit was calculated along with mean solu-
    tion times. Overall, solution rates for homonym analogies
    were at .68 (SD = .26) and solution times averaged 10.54 s
    (SD = 6.07). Table 3 shows the proportion correct per analogy.
    In addition to solution times and rates, we were interested in
    the types of errors participants made on these new homonym

    analogies. In particular, we were interested in which of three
    incorrect choices participants made when they were unable to
    solve the analogy correctly and how quickly they made these
    errors. Errors were categorized into one of three possible types
    for each analogy based on their selection on the multiple-
    choice portion of the task. Errors were defined as correct con-
    text associate errors, incorrect context high associate errors, or
    incorrect context low associate errors (see Table 4).

    A one-way ANOVA for error type (correct context
    associate vs. incorrect context high associate vs. incorrect
    context low associate) was conducted on the proportion of
    each error selected for the analogies. A significant main
    effect was found, with post hoc pairwise comparisons
    revealing that significantly more incorrect context high
    associate errors were made (M = .5, SD = .48) than correct
    context associate errors (M = .14, SD = .33) or incorrect
    context low associate errors (M = .11, SD = .29), with the
    latter two not differing significantly from one another,
    F(2,110) = 14.62, p < .01, η2p = .28.

    For solution times, we examined whether participants
    differed between correctly and incorrectly solved analo-
    gies. A paired t test revealed that solution times for correct
    answers (M = 10.30s, SD = 5.02) were significantly faster
    than solution times for errors (M = 15.00s, SD = 10.27),
    t(41) = -3.25, p < .01.

    Given that the homonyms used in the c position of the
    analogy all have at least two (and sometimes three or four)
    contextual meanings, it is important to consider whether the
    ordinarily dominant context (regardless of the contextual in-
    terpretation of the a and b terms in the analogy) influences the
    types of answers selected, whether correct or incorrect. Using
    the Twilley et al. (1994) norms, we obtained dominance rat-
    ings for the context of each homonym (c term) in terms of
    whether the incorrect analogy context afforded by the a to b
    terms was consistent with the normally dominant

    Table 3 Proportion correct (standard deviations in parenthesis) for each
    homonym analogy

    Analogy (solution) Proportion correct

    run : legs :: stitch : needle .79 (.42)

    table : surgery :: organ : music .36 (.49)

    roar : lion :: horn : car .79 (.42)

    weapon : gun :: bug : spider .88 (.32)

    vowel : letter :: capital : city .83 (.39)

    engrave : wood :: log : book .41 (.51)

    alive : dead :: wake : sleep .95 (.22)

    February : month :: date : fruit .50 (.51)

    starve : eat :: fast : slow .60 (.50)

    Table 4 Proportion of errors (standard deviations in parenthesis) for
    each homonym analogy by error type

    Analogy (solution)

    Correct
    context

    associate

    Incorrect
    context high
    associate

    Incorrect
    context low
    associate

    run : legs :: stitch : needle .36 (.49) .41 (.50) .24 (.43)

    table : surgery :: organ :music .13 (.34) .70 (.46) .17 (.37)

    roar : lion :: horn : car 0 1 (0) 0

    weapon : gun :: bug : spider .41 (.50) .53 (.51) .06 (.24)

    vowel : letter :: capital : city .09 (.30) .81 (.40) .09 (.30)

    engrave : wood :: log : book .15 (.39) .73 (.45) .12 (36)

    alive : dead :: wake : sleep 0 .1 (.57) 0

    February : month :: date : fruit 0 .87 (.33) .12 (.33)

    starve : eat :: fast : slow .52 (.51) .24 (.44) .24 (.44)

    886 Mem Cogn (2015) 43:879–895

    interpretation of the homonym, or whether the analogy answer
    (i.e., the correct context) was consistent with the dominant
    interpretation. Taking the analogies presented in Table 3, for
    six of the nine analogies the context of the correct answer was
    also the dominant context of the homonym (analogy solu-
    tions—spider, slow, needle, city, car, sleep), whereas for the
    remaining three analogies (analogy solutions—fruit, music,
    book) the incorrect analogy context established by the a and
    b terms was the dominant context (or a more dominant con-
    text) than the answer context of the homonym. It was not
    possible to have an even division of homonym analogies in
    which the correct or incorrect context was the dominant con-
    text of the homonym due to the constraints of material design
    to obtain accurate homonym analogies with appropriate DRM
    lists. As such, it was decided that the majority of analogies
    should be within the category where the dominant context was
    the answer context.

    Given this, the question arises as to whether it is easier to
    access the analogy solution when the required context for the
    answer is also known to be the most dominant context of the
    homonym (as determined by norms established by Twilley
    et al., 1994). Although we cannot directly compare across
    responses because participants completed three of the nine
    analogies, it is possible to provide mean solution rates and
    times for the analogies. The mean solution rates and times
    indicated that when the answer to the analogy was consistent
    with the dominant context of the homonym, participants
    seemed to access the answer more readily (M = 10.52 s),
    and with greater accuracy (M = .81) in comparison to when
    the context established by the a and b analogy terms was more
    dominant (M = 12.43 s, M = .43, respectively). This suggests
    that homonym dominance might make the answer easier to
    access, and the a, b, and c terms of the analogy somewhat
    easier to inhibit, if the answer is consistent with the dominant
    homonym interpretation.

    If we take the correct context errors and incorrect context
    high associate errors and look at these in terms of homonym
    dominance, it would seem as if participants made, on average,
    more incorrect context high associate errors when this incor-
    rect context was the dominant context of the homonym (M =
    .71) in comparison to when the answer or correct context was
    dominant (M = .63). If we look at the correct context associate
    errors, participants make more of these particular errors when
    the answer context (correct context) was dominant (M = .16)
    than when the context established by the a and b analogy
    terms (incorrect context) was dominant (M = .08). Of course,
    these effects have to be interpreted with caution, given the
    overall low rate of correct context response errors overall
    compared to incorrect context high associate errors, regardless
    of homonym dominance. However, the fact remains that hom-
    onym dominance can have an influence on response errors
    inasmuch as errors tend to be consistent with the dominant
    interpretation of the homonym.

    Discussion

    The results of Experiment 2a provide evidence that errors
    made while solving homonym analogies are through a bias
    towards selecting the highest semantic associate to the c term
    in the analogy, even when this item is incorrect. When a hom-
    onym analogy was solved incorrectly, participants were sig-
    nificantly more likely to have selected the highest semantic
    associate of the incorrect context (i.e., the semantic associate
    of c interpreting the homonym in the context set by the a and b
    terms of the analogy), rather than a lower associate of the
    incorrect context, or a semantic associate of the correct
    context.

    The tendency toward selecting the highest semantic asso-
    ciate of the c term (in the context established by the a and b
    items) during an error response, suggests a bias towards
    selecting a high semantic associate of c, even when this item
    is not the correct one when solved by an analytic process of
    relational mapping. The propensity to be drawn toward solv-
    ing verbal proportional analogies by semantic association is
    well established, particularly in terms of how children solve
    these analogies (Sternberg & Nigro, 1980; see also Ball et al.,
    2010; Cheshire et al., 2007; Siegler & Svetina, 2002).
    However, it is widely believed that adults utilize a more so-
    phisticated process of relational mapping to arrive at the cor-
    rect answer (Green et al., 2006). In contrast, what the current
    analysis of the errors made during the solving of homonym
    analogies suggests is that adults are also drawn to a high se-
    mantic associate of the c termwhen solving verbal proportion-
    al analogies. One possibility is that those making errors use
    semantic association as a heuristic to aid in selecting the solu-
    tion, rather than identifying the relationship between the initial
    two analogy terms, and applying this to the latter part of the
    analogy. In other words, people may be defaulting to the use
    of semantic association rather than reasoning by means of an
    analytic mapping process based on the relation that exists
    between the a and b terms within the analogy. If this is the
    case, we would expect that participants who are drawn to-
    wards making an incorrect context high or low associate error
    would also solve analogies faster than those who solve the
    analogies correctly.

    Previous standard forms of analogical reasoning problems
    have rendered it difficult, if not impossible, to distinguish
    between the correct solution strategy of relational mapping
    (Green et al., 2006) versus the potentially incorrect method
    of solving by semantic association. This is because typical
    verbal analogical problems are often confounded by the fact
    that the c and d terms not only have a relational link but are
    also often highly semantically associated (Howe et al., 2013).
    For example, in the problem ‘pyramid is to cube as triangle is
    to square,’ triangle and square are highly semantically associ-
    ated, and participants might be likely to generate square in the
    absence of analogical reasoning. The use of our newly

    Mem Cogn (2015) 43:879–895 887

    designed homonym analogies demonstrates that the applica-
    tion of this heuristic by adults can lead participants to arrive at
    an incorrect solution. Indeed, the current findings suggest that
    the context of the analogy is important in participants’ overall
    decision when selecting a solution, such that participants are
    often drawn towards an incorrect answer that fits with the
    context of the homonym established by the a and b terms,
    rather than the alternate context established by the c term.
    Thus, the overall context and the relation to the semantic as-
    sociation of cmight be important in solving typical analogies.

    Furthermore, responses to homonym analogies can be in-
    fluenced by the dominant context of the homonym term.
    However, even when the dominant context is the correct
    answer, it can be difficult to overcome the context set
    by the analogy terms. Homonym dominance can lead to
    errors consistent with the dominant interpretation of the
    homonym. For the majority (six out of the nine) of the
    homonym analogies presented in Experiment 2a, the
    dominant context was the answer context (e.g., the ‘cor-
    rect context’), yet participants were still drawn to mak-
    ing incorrect context associate errors when solving hom-
    onym analogies (and were biased by the context provid-
    ed in the analogy terms), even when this was not the
    dominant context of that homonym. Despite the fact that
    these analogy problems seem to be solved more accu-
    rately when the analogy solution context was the dom-
    inant interpretation, it still did not prevent participants
    from making incorrect context response errors. That is,
    participants’ responses may be biased by the incorrect
    context established by the a and b terms of the analogy
    even when this context is a less dominant than the
    correct context. Given that accuracy rates for the anal-
    ogies in which the correct context is dominant were far
    from ceiling, and that there were significantly more in-
    correct context errors in the set of analogies where this
    incorrect context was not a dominant interpretation,
    homonym dominance does not entirely determine re-
    sponse selection. Indeed, participants were mainly influ-
    enced by the context of the a and b analogy terms
    when interpreting the c term and not simply relying
    on their existing knowledge base of homonym context
    interpretations.

    The use of homonym analogies demonstrates that errors are
    made when participants have a bias to generate the solution
    term in the context of the a and b components of the analogy
    such that they then search for a similar semantic associate of c.
    The correct solution to a homonym analogy—one that in-
    volves the a–b relation—necessitates inhibition of not only
    the context provided by the a and b terms, but also of the
    highest semantic associate of c to this context. Research has
    demonstrated that young children often struggle with inhibi-
    tory control in analogy problems (e.g., Richland, Morrison, &
    Holyoak, 2006). Experiment 2a provided evidence that errors

    made by adult participants in analogical reasoning with hom-
    onym problems can also arise from difficulty in inhibiting the
    context of the analogy and the automatic spreading activation
    to semantic associates of c to this context.

    Experiment 2b

    The aim of Experiment 2b was to ascertain if false memory
    priming can help adults overcome the bias observed in
    Experiment 2a, whereby they tend to generate the solution to
    a proportional analogy by searching for a semantic associate
    of c in the context established by the a and b terms of the
    analogy. Generating a false memory at recall would be expect-
    ed to make this item more salient in memory, thereby priming
    the availability of this item as a solution term during subse-
    quent analogical reasoning. We therefore expected that false
    memory priming would benefit participants in that they would
    be able to inhibit the tendency to use the heuristic in the anal-
    ogy that leads to the incorrect answer, that is, simply generat-
    ing a high semantic associate of c in terms of the (incorrect)
    context that is established by the a and b terms of the analogy.

    Method

    Participants

    A total of 46 females aged 18 years participated in the exper-
    iment. Each participant provided written informed consent
    prior to taking part in the experiment, and participants were
    fully debriefed at the end. All participants were fluent in
    English.

    Design and materials

    We employed a within-participant design similar to
    Experiment 1. This consisted of one factor with two levels
    (Priming: Unprimed or False Memory Primed). The experi-
    ment was programmed using Psyscript and played by an
    Apple Macintosh computer. Thirteen analogies were used in
    this experiment (see Appendix C). Of the 13 problems, 10
    critical analogies were designed such that the c term within
    the analogy was a homonym term (e.g., bark as ‘the noise a
    dog makes’ or, consistent with the a is to b relation in the
    analogy, bark as ‘the outer lining of a tree’). The other three
    analogical problems were included to form distractor items.
    Three distractor items were included in the set of solved ana-
    logical problems to ensure that participants did not identify a
    consistent pattern within the homonym analogies such that the
    answer was always the opposite context to the terms presented
    within the analogy. One of these three was also a homonym
    analogy, while the remaining analogies were non-homonym
    analogies similar to those used in Experiment 1 (see Appendix

    888 Mem Cogn (2015) 43:879–895

    A). The 10 critical analogical problems (see Appendix C)
    were divided into two groups equated on the BAS of the
    DRM list items associated with each analogy problem
    (Group 1: Mean BAS = .33, Group 2: Mean BAS = .24). In
    Group 1, four out of five analogies had the dominant interpre-
    tation of the homonym as the correct context; in Group 2, this
    was three out of five analogies.

    For each analogical problem, participants were presented
    with a choice of four items from which to select their answer.
    Only one of these answers was the correct answer. For the 10
    critical analogies, the three alternative responses were com-
    posed of an associate of the correct answer and two associates
    of an incorrect context answer.

    For each analogical problem there was a linked DRM list
    consisting of 12 associated words where the critical lure was
    the problem solution (refer to Appendix C for the DRM lists
    and the associated BAS scores). DRM-list words that overlap-
    ped with the items presented in the analogical problems were
    removed so that DRM items were not presented as part of any
    subsequent analogy items. The single exception to this was the
    word fast, which was integral to the DRM list slow and was
    therefore left in the list. DRM lists were selected such that they
    only primed one context (refer to Appendix C for DRM lists).
    Presentation of the materials was counterbalanced such that
    each analogy group served in the unprimed and primed
    conditions.

    Procedure

    The procedure was identical to that of Experiment 1 except
    that participants completed 13 analogies rather than eight.
    Furthermore, participants were asked to choose an answer
    from one of four provided. These options were displayed di-
    rectly beneath the analogy, and were labeled a to d. The posi-
    tion of the correct answer was randomized across participants.

    Results

    The mean analogy solution rate (proportion) and the mean
    analogy solution time (seconds) were calculated for each par-
    ticipant. Solution rates and times were analyzed separately in
    two analyses of variance (ANOVAs). For the primed analogy
    problems, solut ion rates and t imes were fur ther
    conditionalized according to whether the participant produced
    the critical lure item (i.e., primed and produced a false mem-
    ory) or did not (i.e., primed but did not produce a false mem-
    ory). Therefore, like Experiment 1, for the purposes of the
    analyses there were three priming conditions (i.e. unprimed
    vs. primed with no false memory vs. primed with false mem-
    ory recalled). Like Experiment 1, the majority of participants
    (over 75 %) contributed data to all three cells. The mean false
    memory proportion was .34 (SD = .21), with the majority of
    participants (78 %) having one or more false memories.

    Solution rates

    There was a significant main effect of priming on solution
    rates, F(2, 58) = 10.3, p < .001, η2p = .26. As can be seen in
    Fig. 3, and which was confirmed using post hoc pairwise
    comparisons, solution rates were significantly higher in the
    false memory condition when a critical lure had been pro-
    duced (M = .89, SE = .05) in comparison to either false mem-
    ory priming where no critical lure was produced during recall
    (M = .69, SE = .04, p < .05) or the unprimed condition (M =
    .62, SE = .05, p .05). Figure 3 displays
    solution rates for each condition.2

    Solution times

    There was no significant main effect of priming on solution
    times, F(2, 56) = .659, p > .05. Analogical problem solutions
    were solved equally fast when unprimed (M = 11.31 s, SE =
    1.03), primed with no false memory (M = 11.87 s, SE = .94),
    or primed with false recall of the critical lure (M = 10.56 s, SE
    = 10.57). However, Mauchly’s test revealed that the assump-
    tion of sphericity had been violated for the solution time data
    (p = .02), indicating that there was considerable variability
    across participants in solution times. Furthermore, examina-
    tion of a histogram suggested that the solution time data were
    bimodally distributed, in that there were two groups of solu-
    tion times, reflecting participants who were fast solvers and
    participants who were slow solvers.

    Because of the bimodal distribution, we decided to exam-
    ine solution times separately for fast solvers and slow solvers
    by splitting participants on the basis of their mean solution
    times for unprimed analogies. This method of splitting
    solution time data into two groups of fast and slow solvers is
    consistent with that described by Garner and Howe (2014)
    when analyzing solution times for CRAT problems. In what
    follows, we describe analyses that included the addition of
    group (fast vs. slow solvers) as a post hoc between-
    participants factor.

    Comparing fast and slow problem solvers It should be noted
    that a one-way ANOVA on solution rates revealed a margin-
    ally significant difference between the fast and slow solvers,
    F(1, 45) = 4.189, p =.047, where fast solvers were slightly less
    accurate (M = .62) than slow solvers (M = .71) in their re-
    sponses. This suggests that fast responding might generate a
    speed accuracy trade-off. For the solution times, a 2 × 3 mixed
    ANOVA (Group x Priming) was conducted with group as the

    2 Solution rates were examined separately for only the analogies in which
    the answer context corresponded with the dominant context of the hom-
    onym. A repeated measures ANOVA revealed that there was no signifi-
    cant main effect of priming on solution rates, F(2, 58) = 1.71, p > .05, for
    these analogies.

    Mem Cogn (2015) 43:879–895 889

    between-participants factors with two levels (fast vs. slow
    solvers) and the within-participant factor of priming with three
    levels (unprimed vs. primed no false memory vs. primed with
    false memory). Not unexpectedly, the results showed that
    there was a significant main effect of group, F(1, 27) =
    323.13, p .05, with problems being solved equally
    quickly in the unprimed (M = 11.49 s, SE = .81), primed with
    no false memory (M = 12.06 s, SE = .78), and primed with
    false memory (M = 10.25 s, SE = 1.00). However, consistent
    with our intuition (and Garner & Howe’s, 2014, previous
    CRAT findings), there was a Group x Priming interaction,
    F(2, 54) = 4.19, p <.05, η2p = .13 (see Fig. 4).3

    To ascertain the source of the significant interaction we
    employed simple main effects analyses using a Bonferroni
    correction for multiple comparisons. There was a significant
    difference between fast and slow solvers in the unprimed con-
    dition (p < .001) such that fast solvers completed the unprimed
    analogical problems significantly faster (M = 8.24 s, SE = .64)
    than the slow solvers (M = 16.32 s, SE = 1.18). Fast solvers (M
    = 9.64 s, SE = 1.01) also solved the problems significantly
    faster than slow solvers (M = 15.52 s, SE = 1.28) when primed
    with no false memory (p < .001). However, there was no
    significant difference between fast solvers (M = 9.91 s, SE =
    1.42) and slow solvers (M = 11.63 s, SE = 1.82) in the primed

    with a false memory condition (p > .05). This indicates that for
    participants who were primed and who produced a false mem-
    ory at recall, the priming effect benefits the slow solvers (i.e.,
    they were as fast as the fast solvers) perhaps because the fast
    solvers are already at ceiling for solution times.

    Discussion

    The findings from Experiment 2b demonstrate that false
    memory priming of homonym analogy problems leads
    to significantly higher solution rates than when those
    same problems are unprimed or are primed but no false
    memories are generated at recall. Moreover, the results
    of Experiments 2a and 2b show that when participants
    make errors in solving homonym analogies they have a
    tendency to opt for a high semantic associate of the
    incorrect context (in other words, the context consistent
    with the a and b analogy terms), but this context bias is
    frequently overcome when priming is effective. These
    findings provide evidence that priming may help partic-
    ipants overcome a bias with selecting the high semantic
    associate consistent with the analogy problem and may
    also increase a participant’s ability to inhibit the context
    set by the a and b terms when interpreting the hom-
    onym. From Experiment 2b it seems that falsely
    recalling a non-presented critical item is linked to a
    more efficient ability to inhibit the incorrect context of
    the analogy and to select the correct context item in
    analogical reasoning. This is consistent with the idea
    that false memory primes are particularly effective at
    priming problem-solving tasks (more so than true
    primes), such that they have the strength to enable in-
    hibition of even a dominant context in problem solving.

    The present findings also extend the efficacy of false
    priming in terms of the time taken to solve analogical
    reasoning problems. Previous research has demonstrated

    Fig. 4 Mean solution times (s) with standard errors for fast and slow
    solvers as a function of priming condition

    3 It is important to note that it was not possible to conduct analyses to
    determine if the type of priming (unprimed vs. primed no false memory
    vs. primed false memory) influenced the type of errors (correct context
    associate vs. incorrect context high associate vs. incorrect context low
    associate) for fast and slow solvers because there were few instances
    where participants were primed (and either produced or did not produce
    a false memory) and did not produce a correct response to the correspond-
    ing analogy item.

    0

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    Unprimed Primed No

    False Memory

    Primed False

    Memory

    M
    ea

    n
    S

    o
    lu

    ti
    o

    n
    R

    at
    e

    Priming

    Fig. 3 Mean proportion of solution rates (with standard errors) as a
    function of priming condition

    890 Mem Cogn (2015) 43:879–895

    that false memory priming results in problems being
    solved more quickly than those that are unprimed, but
    these results are confined to analogies whose solutions
    can be easily generated via spreading activation (Howe
    et al., 2013). What the evidence here suggests is that
    even for problems whose solutions may not be as easily
    generated via spreading activation (at least to near asso-
    ciates) their solutions can also be primed by false mem-
    ories. Moreover, these results show that there are signif-
    icant individual differences in participants’ solution
    times such that a subset of participants complete ana-
    logical reasoning problems with a speed that leaves lit-
    tle room for any possible improvement provided by
    priming (our fast solvers subset). Given that the mean
    solution time for the fast solvers is approximately 8 sec-
    onds, which includes reading the analogy as well as the
    four response items, it is unlikely that the analogies
    could be completed more rapidly than this, leaving little
    room for priming effects. However, there is a subset of
    participants (our slow solvers) where false memory
    priming does improve solution times to levels compara-
    ble to that of fast solvers, demonstrating the efficacy of
    false memory priming in homonym analogy problems.

    General discussion

    Previous research has established that false memories
    can have salutary effects (Howe, 2011; Schacter et al.,
    2011). One positive effect concerns the ability of false
    memories to prime solutions on problem-solving tasks
    involving insight-based reasoning (i.e., CRATs; see
    Howe et al., 2010b, 2011; Garner & Howe, 2014).
    Indeed, false memories have proved to be more effec-
    tive primes for CRAT solutions than true memories
    when a delay (e.g., 1 week) has been imposed between
    the time participants recall words from studied lists and
    the time they are presented with CRAT problems (see
    Howe et al., 2012). Previous research has also demon-
    strated that simple verbal analogy problems can be
    primed using false memories but not true ones (Howe
    et al., 2013).

    However, these findings have been restricted to con-
    ditions in which priming effects may have occurred as a
    result of simple spreading activation through local and
    highly interconnected semantic associates. This is of
    particular concern for studies examining analogical rea-
    soning (Howe et al., 2013) because the problems used
    there may not have required analogical reasoning per se.
    That is, the problems could have been solved using
    simple associations involving BAS from the a, b, and
    c terms to the solution d term. This means that

    participants would not have had to understand the a to
    b relation in order to solve the problem.

    The novel contribution of this series of experiments, ex-
    tending our understanding of the adaptive consequences of
    false memories, is that false memory priming occurs even in
    the absence of obvious associative relations among items.
    That is, the current experiments made it more difficult
    to use only spreading activation through semantic asso-
    ciations to solve analogical reasoning problems by elim-
    inating BAS within the analogy (Experiment 1) or by
    using homonym analogies (Experiments 2a and 2b). The
    findings across these three experiments provide evidence
    that false memories are effective primes for solutions on
    analogical reasoning tasks even when those solutions
    may not rely heavily (or perhaps at all) on spreading
    activation through semantic associative networks. That
    is, we have demonstrated for the first time that self-
    generated false memories can and do prime solutions
    to problem-solving tasks—verbal proportional analo-
    gies—in which the BAS of the analogy terms was lim-
    ited so that participants had to rely on reasoning using
    the a to b relation to generate the c to d solution.

    Furthermore, we have developed a novel set of anal-
    ogies (termed homonym analogies) that require the inhi-
    bition of semantic associates provided by the context of
    the analogy in order to generate a logical (relational)
    solution to the analogical reasoning problem. Although
    existing knowledge based interpretation of homonyms
    might influence responses to these analogies, responses
    are primarily guided by interpreting the homonym in
    terms of the context provided by the a, b, and c terms
    of the analogy. False memory priming of the correct
    solution facilitated participants’ analogical reasoning
    not only in terms of solution rates but also in how
    quickly solutions were achieved (at least for participants
    who were not already at or near ceiling). Interestingly,
    when participants make errors on these analogies they
    do so with items that would have been generated via
    spreading activation through semantic associates. Of
    course, it is premature to conclude that adults’ (like
    children’s) default analogical reasoning heuristic is a
    search through associative networks, ones created by
    the biasing context of the homonym analogy itself.
    However, it is clear that false memory priming facili-
    tates analogical reasoning either through the inhibition
    of this initial biasing context or through refocusing the
    search for a solution to networks related to the false
    memory (or both).

    Indeed, our findings do not rule out the idea that
    adults may still use spreading activation through seman-
    tic associative networks to solve analogical reasoning
    problems. Although the two interpretations of the hom-
    onym (one provided by the false memory that has been

    Mem Cogn (2015) 43:879–895 891

    generated and the other by the biasing context of the
    analogy) may not be compatible inasmuch as they are
    not ‘located’ in the same semantic neighborhoods, both
    interpretations will be active in associative memory at
    the same time. Of course, that two disparate interpreta-
    tions of the same concept are active in memory at the
    same time is not unheard of and in some circumstances
    is fully anticipated (e.g., Brainerd, Wang, & Reyna,
    2013; Nelson et al., 2003, 2013).

    However, the question remains as to how adults rec-
    oncile these two interpretations, inhibit the more recent
    contextual bias from the analogy, and supplant it with
    the solution from the older false memory context. One
    likely possibility is that participants use the a to b re-
    lation to search for an alternative interpretation that is
    the solution to the analogy. Furthermore, it could be
    that the increased activation of a concept, through a
    process such as spreading activation, leads to increased
    fluidity of the concept, enabling its use in solving hom-
    onym analogies regardless of the context or interpreta-
    tion. It is important to note that this increased activation
    and accessibility is limited to situations in which the
    critical lure is produced during recall; those circum-
    stances in which the lure is not falsely recalled produce
    no beneficial priming effect in standard or homonym
    analogies. Critical lures that have been activated during
    DRM list presentation are still above threshold activa-
    tion levels and remain highly active in memory when
    participants are solving the analogies, making them
    more accessible as a solution. Therefore, when
    performing a search for an alternative interpretation for
    a homonym analogy, false memory priming works be-
    cause this alternative interpretation of the homonym is
    already active in memory, making this search process
    less difficult. Critical lures that were activated but not
    falsely recalled are thought to have either dropped be-
    low the activation threshold required for priming after
    being rejected or inhibited during test, or to have not
    been activated sufficiently above the threshold required
    for priming during study, thus reducing their accessibil-
    ity when interpreting the meaning of a homonym
    analogy.

    Whether homonym analogies are solved by applying
    the a to b relation to the c term to the generate d term
    or by searching the two distant neighborhoods of se-
    mantic associates that were recently activated for the
    homonym (or, indeed, by some combination of both)
    must await further research. However, the importance
    of the current results is that regardless of whether adults
    use semantic search, analytic mapping processes, or
    both, false memories can have some very positive ef-
    fects inasmuch as they provide a powerful priming
    mechanism for solving problems.

    These findings are not just important from a theoret-
    ical perspective; they carry with them some interesting
    everyday ramifications. This is because, as mentioned
    earlier, false memories occur frequently in a number of
    different contexts, both in and out of the laboratory (see
    Brainerd & Reyna, 2012; Howe, 2013). Often, these
    semantically activated false memories arise spontaneous-
    ly and automatically, outside of the rememberer’s con-
    scious awareness. In the world outside of the laboratory,
    perhaps the best known consequences of these false
    memories are those that have given rise to courtroom
    allegations of offences that may never have occurred
    (Howe, 2013; Schacter & Loftus, 2013). Indeed, as
    shown in the current research, false memories can and
    do serve as the basis for reasoning and decision making,
    and arguably do so not just in the laboratory context
    but in any number of everyday contexts. For example,
    continuing with the forensic theme, it is not just com-
    plainants’ false memories that can lead to decisions to
    prosecute; jurors’ false memories can lead to potential
    miscarriages of justice. Seminal work by Pennington
    and Hastie (1986, 1990, 1991, 1992) has shown that
    jurors activate story schemas based on their attempts
    to understand and integrate trial evidence. These
    schemas not only serve an organizing function but also
    serve to bias additional pieces of evidence as the trial
    proceeds. Worse, jurors can form false, schema-
    consistent memories for Bfacts^ that are not actually in
    evidence. Because jury deliberations involve reasoning
    from such (biased) evidence, decision making as to a
    complainant’s guilt or innocence will be influenced not
    just by correct recollections of the evidence but also by
    (false) memories of facts not in evidence, ones that
    were semantically activated when the juror’s story sche-
    ma was invoked.

    The role of false memories in jury decision making is made
    more ominous given that trials usually involve considerable
    negative emotional content (e.g., sadness, anger, fear; see
    Nuñez, Schweitzer, Chai, & Myers, in press). The evidence
    reviewed earlier (e.g., Howe et al., 2010a) shows that negative
    false memories not only persist over time but can also increase
    over a retention interval (e.g., during the course of a trial). This
    is thought to be due to negative information beingmore dense-
    ly interrelated than other types of information (Talmi, Luk,
    McGarry, & Moscovitch, 2007), which in turn makes spread-
    ing activation more likely through such associative networks.
    Indeed, previous laboratory-based research has shown that
    negative false memories serve as better primes than neutral
    false memories during an insight-based problem-solving ex-
    ercise (e.g., Garner & Howe, 2014).

    What these observations suggest is that because false mem-
    ories can play a role in everyday cognition, including reason-
    ing and decision making, there is a need to study their

    892 Mem Cogn (2015) 43:879–895

    influence, both in controlled laboratory conditions as well as
    in more naturalistic settings. Indeed, studies have shown that
    false memories not only serve as powerful primes in children’s
    and adults’ reasoning tasks (e.g., Howe et al., 2011, 2013),
    some of which are used to assess intelligence and creativity,
    but they also play a key role in tasks frequently used to assess
    more perceptual components of intelligence (e.g., perceptual
    closure tasks; see Otgaar, Howe, van Beers, van Hoof, &
    Bronzwaer, 2015). Understanding the pivotal role false mem-
    ories play in remembering the past, interpreting the present,
    and planning for the future (see Howe, 2011; Schacter et al.,
    2011) is essential if we are to have a complete picture of the
    importance of memory in everyday cognition.

    Acknowledgments This research was supported by a project grant
    from the Economic and Social Research Council, ESRC grant RES-
    062-23-3327.

    Appendix A

    Experiment 1: Analogies and Associated DRM Lists

    Group A Analogies and DRM Lists (with BAS)

    wash : clean :: press : iron
    car : roundabout :: moon : earth
    peace : dove :: courage : lion
    egg : yolk :: plum : stone
    iron: ore, steel, metal, crease, starch, steam, wrinkle, rust,
    copper, calcium, element, magnet (.10)
    earth: planet, world, geology, ground, gravity, environment,
    worm, heaven, sphere, globe, core, atmosphere (.15)
    lion: tiger, circus, jungle, tamer, den, cub, Africa, mane, cage,
    feline, roar, fierce (.16)
    stone: pebble, rock, granite, kidney, sapphire, gem, brick,
    statue, marble, gravel, stick, tomb (.14)

    Group B Analogies and DRM Lists (with BAS)

    leopard : spots :: chest : hair
    four : cat :: eight : spider
    watch : cog :: compass : needle
    prevent : restrict :: enable : allow
    hair: strand, brush, scalp, lice, conditioner, comb, shampoo,
    headband, dandruff, mousse, bald, clippers (.31)
    spider: web, insect, bug, fright, fly, arachnid, crawl, tarantula,
    poison, bite, creepy, feelers (.19)
    needle: thread, pink, eye, sewing, sharp, point, prick, thimble,
    haystack, thorn, hurt, injection (.20)
    allow: permit, let, permission, forbid, disallow, forbidden,
    prohibit, accept, admit, ban, admission, deny (.10)

    Appendix B

    Appendix C

    Experiment 2b Analogical Problems and DRM Lists

    Group A Analogies (with multiple-choice A, B, and C errors,
    respectively) and DRM Lists (with BAS)

    weapon : gun :: bug : spider (phobia, spy, deception)
    starve : eat :: fast : slow (still, food, snack)
    February : month :: date : fruit (cocktail, schedule, calendar)
    run : legs :: stitch : needle (cloth, exercise, pain)
    engrave : wood :: log : book (story, tree, branch)
    spider: web, insect, bug, fright, fly, arachnid, crawl, tarantula,
    poison, bite, creepy, feelers (.19)
    slow: fast, lethargic, stop, listless, snail, cautious, delay, traffic,
    turtle, speed, wait, sluggish (.17)
    fruit: apple, vegetable, orange, kiwi, citrus, ripe, pear, banana,
    berry, cherry, basket, juice (.25)
    needle: thread, pink, eye, sewing, sharp, point, prick, thimble,
    haystack, thorn, hurt, injection (.20)
    book: text, library, chapter, novel, publisher, author, literature,
    reader, page, magazine, read, title. (.52)

    Group B Analogies (with multiple-choice A, B, and C errors,
    respectively) and DRM Lists (with BAS)

    vowel : letter :: capital : city (address, alphabet, number)
    arithmetic : calculator :: rule : king (castle, measure, depth)
    table : surgery :: organ : music (piano, heart, donor)

    Table 5 Analogies normed in Experiment 2a

    Analogy (solution) MC foils

    Correct
    context

    Incorrect
    context high
    associate

    Incorrect
    context low
    associate

    run : legs :: stitch : needle cloth exercise pain

    table : surgery :: organ : music piano heart donor

    roar : lion :: horn : car race rhino Africa

    weapon : gun :: bug : spider phobia spy deception

    vowel : letter :: capital : city address alphabet number

    engrave : wood :: log : book story tree branch

    alive : dead :: wake : sleep peace funeral grave

    February : month :: date : fruit cocktail calendar schedule

    starve : eat :: fast : slow still food snack

    Note. MC = Multiple choice.

    Mem Cogn (2015) 43:879–895 893

    roar : lion :: horn : car (race, rhino, Africa)
    alive : dead :: wake : sleep (peace, funeral, grave)
    city: town, crowded, state, slum, streets, subway, country,
    New York, village, metropolis, big, Chicago (.17)
    king: queen, England, crown, prince, George, dictator, palace,
    throne, chess, subjects, monarch (.25)
    music: sound, harp, sing, radio, band, melody, stereo, concert,
    instrument, symphony, jazz, rhythm. (.24)
    car: truck, bus, train, automobile, vehicle, drive, jeep, Ford,
    keys, garage, highway, van (.35)
    sleep: rest, awake, bed, tired, dream, snooze, blanket, doze,
    slumber, snore, nap, yawn, drowsy (.46)
    Non-critical distractor analogies:
    (non-homonym distractor) caterpillar : tadpole :: butterfly :
    frog (cocoon, monarch, moth)
    (non-homonym distractor) weep : sad :: laugh : happy (cheer,
    joke, silly)
    (homonym distractor) fur : bear :: bark : tree (leaf, orchard,
    log)

    Open Access This article is distributed under the terms of the Creative
    Commons Attribution License which permits any use, distribution, and
    reproduction in any medium, provided the original author(s) and the
    source are credited.

    References

    Ball, L. J., Hoyle, A.M., & Towse, A. S. (2010). The facilitatory effect of
    negative feedback on the emergence of analogical reasoning abili-
    ties. British Journal of Developmental Psychology, 28, 583–602.

    Bassok,M., &Holyoak, K. (1989). Interdomain transfer between isomor-
    phic topics in algebra and physics. Journal of Experimental
    Psychology: Learning, Memory, and Cognition, 15, 153–166.

    Bjorklund, D. F. (2004).Children’s thinking: Development and individual
    differences (4th ed.). Belmont, CA: Wadsworth/Thompson.

    Brainerd, C. J., & Reyna, V. F. (2012). Reliability of children’s testimony
    in an era of developmental reversals. Developmental Review, 32,
    224–267.

    Brainerd, C. J., Reyna, V. F., & Brandse, E. (1995). Are children’s false
    memories more persistent than their true memories? Psychological
    Science, 6, 359–364.

    Brainerd, C. J., Wang, Z., & Reyna, V. F. (2013). Superposition of epi-
    sodic memories: Overdistribution and quantum models. Topics in
    Cognitive Science, 5, 773–799.

    Cheshire, A., Muldoon, K. P., Francis, B., Lewis, C. N., & Ball, L. J.
    (2007). Modelling change: New opportunities in the analysis of
    microgenetic data. Infant and Child Development, 16, 119–134.

    Coltheart, M. (1981). The MRC Psycholinguistic database. Quarterly
    Journal of Experimental Psychology, 33A, 497–505.

    Deese, J. (1959). Influence of interitem associative strength upon imme-
    diate free recall. Psychological Reports, 5, 305–312.

    Garner, S. R., & Howe, M. L. (2014). False memories from survival
    processing make better primes for problem-solving. Memory, 22,
    9–18.

    Gentner, D., & Toupin, C. (1986). Systematicity and surface similarity in
    the development of analogy. Cognitive Science, 10, 277–300.

    Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical
    transfer. Cognitive Psychology, 15, 1–38.

    Goswami, U. (2001). Analogical reasoning in children. In D. Gentner, K.
    J. Holyoak, & B. N. Kokinov (Eds.), The analogical mind:
    Perspectives from cognitive science (pp. 437–470). Cambridge,
    MA: MIT Press.

    Goswami, U., & Brown, A. L. (1989). Melting chocolate and melting
    snowmen: Analogical reasoning and causal relations.Cognition, 35,
    69–95.

    Goswami, U., & Brown, A. L. (1990). Higher-order structure and rela-
    tional reasoning: Contrasting analogical and thematic relations.
    Cognition, 36, 207–226.

    Green, A. E., Fugelsang, J. A., & Dunbar, K. N. (2006). Automatic
    activation of categorical and abstract analogical relations in analog-
    ical reasoning. Memory & Cognition, 34, 1414–1421.

    Gulan, T., & Valerjev, P. (2010). Semantic and related types of priming as
    a context in word recognition. Review of Psychology, 17, 53–58.

    Holyoak, K. J. (2012). Analogy and relational reasoning. In K. J. Holyoak
    & R. G. Morrison (Eds.), The Oxford handbook of thinking and
    reasoning. Oxford, England: Oxford University Press.

    Howe,M. L. (2005). Children (but not adults) can inhibit false memories.
    Psychological Science, 16, 927–931.

    Howe, M. L. (2011). The adaptive nature of memory and its illusions.
    Current Directions in Psychological Science, 20, 312–315.

    Howe, M. L. (2013). Memory development: Implications for adults
    recalling childhood experiences in the courtroom. Nature Reviews
    Neuroscience, 14, 869–876.

    Howe, M. L., Candel, I., Otgaar, H., Malone, C., & Wimmer, M. C.
    (2010a). Valence and the development of immediate and long-term
    false memory illusions. Memory, 18, 58–75.

    Howe, M. L., Garner, S. R., Charlesworth, M., & Knott, L. M. (2011). A
    brighter side to memory illusions: False memories prime children’s
    and adults’ insight-based problem solving. Journal of Experimental
    Child Psychology, 108, 383–393.

    Howe, M. L., Garner, S. R., Dewhurst, S. A., & Ball, L. J. (2010b). Can
    false memories prime problem solutions? Cognition, 117, 176–181.

    Howe, M. L., Threadgold, E., Garner, S. R., Bland, C. E., & Ball, L. B.
    (2015). The development of children’s and adults’ analogical rea-
    soning. Manuscript in preparation.

    Howe, M. L., Threadgold, E., Norbury, J. V., Garner, S., & Ball, L. J.
    (2013). Priming children’s and adults’ analogical problem solutions
    with true and false memories. Journal of Experimental Child
    Psychology, 116, 96–103.

    Howe, M. L., Wilkinson, S., & Monaghan, P. (2012). False memories
    trump true ones as problem-solving primes after a delay.
    Minneapolis, MN: Paper presented at the annual meeting of the
    Psychonomic Society.

    Kahneman, D., & Klein, G. (2009). Conditions for intuitive expertise: A
    failure to disagree. American Psychologist, 64, 515–526.

    Keane, M. (1987). On retrieving analogues when solving problems.
    Quarterly Journal of Experimental Psychology, 39A, 29–41.

    Kokinov, B. (1990). Associative memory-based reasoning: Some exper-
    imental results. In Proceedings of the twelfth annual conference of
    the Cognitive Science Society (pp. 741–749). Hillsdale, NJ:
    Erlbaum.

    Kokinov, B. N., & Petrov, A. A. (2001). Integrating memory and reason-
    ing in analogy-making: The AMBR model. In D. G. Gentner, K. J.
    Holyoak, & B. N. Kokinov (Eds.), The analogical mind:
    Perspectives from cognitive science (pp. 59–124). Cambridge,
    MA: MIT Press.

    McDermott, K. B. (1996). The persistence of false memories in list recall.
    Journal of Memory and Language, 35, 212–230.

    Mednick, M. T., Mednick, S. A., & Mednick, E. V. (1964). Incubation of
    creative performance and specific associative priming. Journal of
    Abnormal and Social Psychology, 69, 84–88.

    894 Mem Cogn (2015) 43:879–895

    Nelson, D. L., Kitto, K., Galea, D., McEvoy, C. L., & Bruza, P. D. (2013).
    How activation, entanglement, and searching a sematic network
    contribute to event memory. Memory & Cognition, 41, 797–819.

    Nelson, D. L., McEvoy, C. L., & Pointer, L. (2003). Spreading activation
    or spooky activation at a distance? Journal of Experimental
    Psychology: Learning, Memory, and Cognition, 29, 42–52.

    Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University
    of South Florida word association, rhyme, and word fragment
    norms. Retrieved from http://w3.usf.edu/FreeAssociation/

    Nuñez, N., Schweitzer, K., Chai, C. A., & Myers, B. (in press). Negative
    emotions felt during trial: The effect of fear, anger, and sadness on
    juror decision making. Applied Cognitive Psychology.

    Newman, E. J., & Lindsay, D. S. (2009). False memories: What the hell
    are they for? Applied Cognitive Psychology, 23, 1105–1121.

    Osman, M., & Stavy, R. (2006). Development of intuitive rules:
    Evaluating the application of the dual-system framework to under-
    standing children’s intuitive reasoning. Psychonomic Bulletin &
    Review, 13, 935–953.

    Otgaar, H., Howe, M. L., Peters, M., Sauerland, M., & Raymaekers, L.
    (2013). Developmental trends in different types of spontaneous false
    memories: Implications for the legal field. Behavioral Sciences and
    the Law, 31, 666–682.

    Otgaar, H., Howe, M. L., Peters, M., Smeets, T., &Moritz, S. (2014). The
    production of spontaneous false memories across childhood.
    Journal of Experimental Child Psychology, 121, 28–41.

    Otgaar, H., Howe, M. L., van Beers, J., van Hoof, R., & Bronzwaer, N.
    (2015). The positive ramifications of false memories using a percep-
    tual closure task. Journal of Applied Research in Memory and
    Cognition, 4, 43–50.

    Otgaar, H., Smeets, T., & Peters, M. (2012). Children’s implanted false
    memories and additional script knowledge. Applied Cognitive
    Psychology, 26, 709–715.

    Pennington, N., & Hastie, R. (1986). Evidence evaluation in complex
    decision making. Journal of Personality and Social Psychology,
    51, 242–258.

    Pennington, N., & Hastie, R. (1990). Practical implications of psycholog-
    ical research on juror and jury decision making. Personality and
    Social Psychology Bulletin, 16, 90–105.

    Pennington, N., & Hastie, R. (1991). A cognitive theory of juror decision
    making: The story model. Cardozo Law Review, 13, 519–557.

    Pennington, N., & Hastie, R. (1992). Explaining the evidence: Tests of
    the story model for juror decision making. Journal of Personality
    and Social Psychology, 62, 189–206.

    Richland, L. E., Morrison, R. G., & Holyoak, K. J. (2006). Children’s
    development of analogical reasoning: Insights from scene analogy
    problems. Journal of Experimental Child Psychology, 94, 249–273.

    Richland, L. E., Zur, O., & Holyoak, K. J. (2007). Cognitive supports for
    analogies in the mathematics classroom. Science, 316, 1128–1129.

    Roediger, H. L., III, & McDermott, K. B. (1995). Creating false memo-
    ries: Remembering words not presented in lists. Journal of
    Experimental Psychology: Learning, Memory, and Cognition, 21,
    803–814.

    Roediger, H. L., III, Watson, J. M., McDermott, K. B., & Gallo, D. A.
    (2001). Factors that determine false recall: A multiple regression
    analysis. Psychonomic Bulletin & Review, 8, 385–407.

    Schacter, D. L., Gallo, D., & Kensinger, E. (2007). The cognitive neuro-
    sciences of implicit and false memories: Perspectives on processing
    specificity. In J. S. Nairne (Ed.), The foundations of remembering:
    Essays in honor of Henry L. Roediger III (pp. 353–378). New York,
    NY: Psychology Press.

    Schacter, D. L., Guerin, S. A., & St. Jacques, P. L. (2011). Memory
    distortion: An adaptive perspective. Trends in Cognitive Sciences,
    15, 467–474.

    Schacter, D. L., & Loftus, E. F. (2013). Memory and law: What can
    cognitive neuroscience contribute? Nature Neuroscience, 16, 119–
    123.

    Schunn, C., & Dunbar, K. (1996). Priming, analogy, and awareness in
    complex reasoning. Memory & Cognition, 24, 271–284.

    Siegler, R. S., & Svetina, M. (2002). A microgenetic/cross-sectional
    study of matrix completion: Comparing short-term and long-term
    change. Child Development, 73, 793–809.

    Slamecka, N. J., & Graf, P. (1978). The generation effect: Delineation of a
    phenomenon. Journal of Experimental Psychology: Human
    Learning and Memory, 4, 592–604.

    Sloman, S. A. (1996). The empirical case for two systems of reasoning.
    Psychological Bulletin, 119, 3–22.

    Sternberg, R. J. (1977). Component processes in analogical reasoning.
    Psychological Review, 84, 353–378.

    Sternberg, R. J., & Nigro, G. (1980). Developmental patterns in the so-
    lution of verbal analogies. Child Development, 50, 27–38.

    Talmi, D., Luk, B. T. C., McGarry, L. M., & Moscovitch, M.
    (2007). The contribution of relatedness and distinctiveness to
    emotionally-enhanced memory. Journal of Memory and
    Language, 56, 555–574.

    Twilley, L. C., Dixon, P., Taylor, D., & Clark, K. (1994). University of
    Alberta norms of relative meaning frequency for 566 homographs.
    Memory & Cognition, 22, 111–126.

    Wilkinson, S. (2014). The false memory illusion strikes back: An
    investigation into the adaptive functions of false memories
    (Unpublished doctoral dissertation). Lancaster, UK: Lancaster
    University.

    Mem Cogn (2015) 43:879–895 895

    http://w3.usf.edu/FreeAssociation/

      Priming analogical reasoning with false memories

      Abstract

      Experiment 1

      Method

      Participants

      Design and materials

      Procedure

      Results

      Solution rates

      Solution times

      Discussion

      Experiment 2a

      Method

      Participants

      Design

      Materials and procedure

      Results

      Discussion

      Experiment 2b

      Method

      Participants

      Design and materials

      Procedure

      Results

      Solution rates

      Solution times

      Discussion

      General discussion

      Appendix A

      Experiment 1: Analogies and Associated DRM Lists

      Group A Analogies and DRM Lists (with BAS)

      Group B Analogies and DRM Lists (with BAS)

      Appendix B

      Appendix C

      Experiment 2b Analogical Problems and DRM Lists

      Group A Analogies (with multiple-choice A, B, and C errors, respectively) and DRM Lists (with BAS)

      Group B Analogies (with multiple-choice A, B, and C errors, respectively) and DRM Lists (with BAS)

      References

    PSYC 575

    Journal Article Summary Assignment Instructions

    Article review. You can select any article pertaining to cognitive psychology, but I suggest that you select an article that you could use in your paper on false memories.  The article should be a peer reviewed journal article that discusses an experiment the authors actually conducted (i.e. a primary source).  In other words, the article cannot be a review article, meta-analysis, editorial or something from the popular press.  In all of these examples the author is discussing someone else’s work and you’re reading about it second-hand.

    Overview

    For each Journal Article Summary, you will choose an article to review and use the
    Journal Article Summary Template to complete the assignment. The article you select must be a peer-reviewed journal article in the field of cognitive psychology. The article must also be a primary source, meaning that the authors are discussing their own research, not others’ research (e.g. review articles). Do not use an article that conducted a meta-analysis. It is ideal to select an article that you will be using in your paper; however, this is not a requirement. If you use an article that does not meet these criteria, you will not receive credit for this assignment. See the example below for detailed explanation of the required material for each question. Submit the completed form and a PDF of the article being reviewed. All material must be in current APA format.

    Instructions

    1. APA reference of article being reviewed

    Write the reference for the article as if it were in the reference section of your paper.

    2. What is the research problem that is being investigated? What is the purpose of the research being conducted?

    Provide the “why” behind the paper. Why have they conducted this experiment? For example: “These experiments were designed to explore the role of second order conditioning in anxiety disorders.”

    3. What is the research question?

    The research question is more specific. What is the specific question or questions the article will answer as a result of the study or experiment? For example: “Are adolescents more sensitive to the memory imparting effects of alcohol?”

    4. What are 2 or more theories that are discussed in the Introduction? How are they used to motivate (or set up) the research question? Do the authors agree or disagree with these theories?

    Simply restate the theories discussed in the introduction in your own words. State how these theories are driving the research questions. If the authors’ hypothesis is correct, will it support the theory or be inconsistent with the theory? You should have good idea of where the authors stand based on the evidence presented and the arguments they are making.

    5. How is the research question operationalized? First, identify the abstract constructs being studied. Next identify the concrete way these are being observed or measured. This should include your IV and DV.

    A construct is an abstract explanatory variable that this not directly observable (e.g. memory). The concrete way the construct is measured will point you to the dependent variable (DV). For example, if the paper is concerned with memory, the DV may be the number of items recalled. The independent variable (IV) could be the amount of sleep each participant was allowed the night before the test. Remember that we cannot directly measure many of the constructs that are studied in psychology, so it is important that we identify how they are being operationalize in each research study.

    6. What is the research design (i.e. between or within subjects, what type of statistical tests were used, what were the levels of each variable)?

    This information will be in the methods section of your paper. Be sure to provide enough detail to describe how the study was designed.

    7. Describe the results (but not their broader implications). Were the results significant? Which ones? Do these support or not support the hypothesis?

    Describe the result in your own words. For example: Group X were able to recall significantly more words than Group Y. This finding supports the hypothesis that manipulation Y would reduce recall.

    8. What limitations are mentioned? Why are these limitations theoretically interesting?

    Limitations can be found in the discussion section of the paper. If a limitation is that they didn’t have X control group, then explain in your own words why that is important. Does it change the interpretation of the findings?

    Note: Your assignment will be checked for originality via the Turnitin plagiarism tool.

    Page 2 of 2

    PSYC 575

    Journal Article Summary Template

    Answer the following 8 questions and submit the completed
    Journal Article Summary Template as well as a PDF of the article being reviewed to Canvas.

    1. Current APA reference of article being reviewed:

    2. What is the research problem that is being investigated? What is the purpose of the research being conducted?

    3. What is the research question?

    4. What are 2 or more theories that are discussed in the Introduction? How are they used to motivate (or set up) the research question? Do the authors agree or disagree with these theories?

    5. How is the research question operationalized? First, identify the abstract constructs being studied. Next, identify the concrete way these are being observed or measured. This should include your IV and DV.

    6. What is the research design (i.e. between or within subjects, what type of statistical tests were used, what were the levels of each variable)?

    7. Describe the results (but not their broader implications). Were the results significant? Which ones? Do these support or not support the hypothesis?

    8. What limitations are mentioned? Why are these limitations theoretically interesting?

    Note: Your assignment will be checked for originality via the Turnitin plagiarism tool.

    Expert paper writers are just a few clicks away

    Place an order in 3 easy steps. Takes less than 5 mins.

    Calculate the price of your order

    You will get a personal manager and a discount.
    We'll send you the first draft for approval by at
    Total price:
    $0.00