Do Emotional Images Improve Memory for Paired Content under Reward Conditions?

16
Do Emotional Images Improve Memory for Paired Content under Reward Conditions? James Laurie Abstract Emotion increases memory for central features of emotional events. Emotional stimuli have also been observed to interfere with the processing and memory of proximate information. However, no consensus has emerged about what constitutes the central feature of an emotional event and whether different processing strategies are used for different types of emotional events. Reward has also been shown to increase performance in a memory task. Recent theories of emotion have stressed the close relationship between reward and emotion. It is surprising then, that few experiments have explored the relationship between the two during cognitive processes such as memory tasks. The present study used a word recognition task to explore whether presenting neutral words with semantically unrelated emotional images would effect memory for the words in either direction. Half the participants were incentivised with a reward to explore whether reward would increase performance in the memory task and to investigate the interaction between reward and positive affect. No significant differences were found in either condition and no interaction was observed. Results suggest that pairing neutral information with unrelated positive emotional stimuli does not increase memory for the neutral information. Results may suggest that task focus and goal priorities can eliminate the interference effects of emotional stimuli. Finally, although further is needed, results may also suggest that memory advantages for information encoded in positive emotional situations require information to be semantically or contextually related to the emotional event. Introduction Emotional events have repeatedly been shown to be remembered more and with greater clarity than neutral events. This has been shown across a wide variety of paradigms including analysis of emotional autobiographical memories (Talirico and Rubin, 2003, Talirico, et al 2009), experimentally controlled studies of episodic memory (Christianson and Loftus, 1991) and comparisons of memory for emotional versus neutral words (LaBar & Phelps, 1998; Kensinger & Corkin, 2003) and pictures (Ochsner, 2000; Hamann et al 1999) using both recognition and recall tasks. Some experiments have found evidence for Easterbrook’s (1959) cueutilization hypothesis, which states that emotional arousal leads to attentional narrowing and therefore memory impairment for information presented with emotional stimuli. For instance Christianson and Loftus, (1991) found that core features of complex emotional scenes were remembered with greater clarity at the expense of peripheral information. Kensinger et al (2007) had people view either emotionally

Transcript of Do Emotional Images Improve Memory for Paired Content under Reward Conditions?

Do  Emotional  Images  Improve  Memory  for  Paired  Content  under  Reward  Conditions?  

 James  Laurie  

   Abstract    Emotion  increases  memory  for  central  features  of  emotional  events.    Emotional  stimuli  have  also  been  observed  to  interfere  with  the  processing  and  memory  of  proximate  information.  However,  no  consensus  has  emerged  about  what  constitutes  the  central  feature  of  an  emotional  event  and  whether  different  processing  strategies  are  used  for  different  types  of  emotional  events.    Reward  has  also  been  shown  to  increase  performance  in  a  memory  task.    Recent  theories  of  emotion  have  stressed  the  close  relationship  between  reward  and  emotion.    It  is  surprising  then,  that  few  experiments  have  explored  the  relationship  between  the  two  during  cognitive  processes  such  as  memory  tasks.    The  present  study  used  a  word  recognition  task  to  explore  whether  presenting  neutral  words  with  semantically  unrelated  emotional  images  would  effect  memory  for  the  words  in  either  direction.    Half  the  participants  were  incentivised  with  a  reward  to  explore  whether  reward  would  increase  performance  in  the  memory  task  and  to  investigate  the  interaction  between  reward  and  positive  affect.    No  significant  differences  were  found  in  either  condition  and  no  interaction  was  observed.    Results  suggest  that  pairing  neutral  information  with  unrelated  positive  emotional  stimuli  does  not  increase  memory  for  the  neutral  information.    Results  may  suggest  that  task  focus  and  goal  priorities  can  eliminate  the  interference  effects  of  emotional  stimuli.    Finally,  although  further  is  needed,  results  may  also  suggest  that  memory  advantages  for  information  encoded  in  positive  emotional  situations  require  information  to  be  semantically  or  contextually  related  to  the  emotional  event.    Introduction    Emotional  events  have  repeatedly  been  shown  to  be  remembered  more  and  with  greater  clarity  than  neutral  events.    This  has  been  shown  across  a  wide  variety  of  paradigms  including  analysis  of  emotional  autobiographical  memories  (Talirico  and  Rubin,  2003,  Talirico,  et  al  2009),  experimentally  controlled  studies  of  episodic  memory  (Christianson  and  Loftus,  1991)  and  comparisons  of  memory  for  emotional  versus  neutral  words  (LaBar  &  Phelps,  1998;  Kensinger  &  Corkin,  2003)  and  pictures  (Ochsner,  2000;  Hamann  et  al  1999)  using  both  recognition  and  recall  tasks.        Some  experiments  have  found  evidence  for  Easterbrook’s  (1959)  cue-­‐utilization  hypothesis,  which  states  that  emotional  arousal  leads  to  attentional  narrowing  and  therefore  memory  impairment  for  information  presented  with  emotional  stimuli.    For  instance  Christianson  and  Loftus,  (1991)  found  that  core  features  of  complex  emotional  scenes  were  remembered  with  greater  clarity  at  the  expense  of  peripheral  information.    Kensinger  et  al  (2007)  had  people  view  either  emotionally  

arousing  objects  or  emotionally  neutral  objects  against  neutral  scenes.    As  expected,  people  showed  greater  memory  for  the  arousing  objects  than  the  neutral  objects.    They  showed  worse  memory,  however,  for  information  in  the  background  scenes  of  the  emotionally  arousing  objects  than  in  the  background  scenes  behind  neutral  objects.    The  emotional  stimuli  had  appeared  to  impair  memory  for  peripheral  information.        The  emotional  stroop  task  experimental  paradigm  has  explored  the  effect  of  emotion  on  cognitive  processes  including  attention  and  control.      This  variation  of  the  classic  stroop  interference  task  presents  participants  with  neutral  or  emotional  /  taboo  words  (i.e.  kill)  presented  in  different  colours.      The  paradigm  has  consistently  found  that  emotional  words  interfere  with  performance  in  the  stroop  task;  emotional  words  are  responded  to  slower  and  with  less  accuracy.          (Siegrist,  1995,  Whalen  et  al.,  1998,).        However,  while  negative  emotional  states  have  been  shown  to  narrow  attention  on  the  source  of  arousal,  positive  emotional  states  have  been  shown  to  broaden  processing  and  increase  creativity  (Fredrickson,  2001).    Furthermore,  participants  are  not  passive  responders  to  stimuli  as  Eastebrook’s  (1958)  cue  utilisation  hypothesis  suggests,  particularly  if  they  are  given  a  goal  to  prioritise,  such  as  the  task  of  memorizing  words.    Easterbrook  limits  behavior  to  an  automatic  response  to  the  source  of  arousal.    By  contrast,  Levine  and  Edelstien,  2010  offers  a  more  nuanced  consideration  of  the  effects  of  emotion  on  memory  that  allows  for  a  complex  interaction  between  top-­‐down  processing  of  goal-­‐relevant  information  and  bottom-­‐up  stimulus-­‐response.        Other  studies  have  found  evidence  for  memory  enhancement  of  incidental,  peripheral  and  proximate  information  of  emotional  scenes.    For  instance,  Libkuman  et  al  1990  found  that  memory  for  both  central  and  irrelevant,  peripheral  aspects  of  a  scene  was  enhanced  during  emotional  arousal.    Information  that  is  temporarily  proximate  has  been  shown  to  receive  processing  priority:  A  study  by  Anderson  et  al  (2006)  found  that  memory  for  neutral  pictures  was  enhanced  if  they  were  presented  temporarily  immediately  prior  or  immediately  after  unrelated  arousing  emotional  images.    Mather  and  Nesmith  (2008)  demonstrated  that  spatial  information  for  emotional  stimuli  is  prioritized  in  processing.    They  found  that  participants  demonstrated  increased  memory  for  the  location  of  affective  and  arousing  pictures  presented  on  a  computer  screen  when  compared  with  neutral  pictures.    Doerksen  and  Shimamura  (2001)  found  a  memory  advantage  for  the  colour  of  previously  presented  affective  words  over  neutral  words.          Information  that  is  contextually  and  semantically  relevant  to  the  emotional  aspect  of  the  scene  also  appears  to  receive  priority  in  processing  for  memory.    This  includes  both  semantic  context  and  social  context.    For  instance,  studies  by  Brierley  et  al  (2007)  and  Guillet  &  Arndt  (2010)  found  that  non-­‐target  words  embedded  in  sentences  with  emotional  target  words  were  remembered  more  than  the  non-­‐target  words  embedded  in  otherwise  matched  neutral  sentences.    Increased  memory  for  peripheral  information  is  also  found  where  that  information  is  processed  social  

contexts.    Studies  by  Tsukiura  &  Cabezaa  (2008)  and  Bridge  et  al  (2010)  demonstrated  that  the  retrieval  of  face-­‐name  associations  was  more  accurate  and  faster  for  smiling  than  neutral  faces.    The  above  studies  show  evidence  that  neutral  information  processed  in  semantic,  temporal  or  spatial  proximity  to  an  emotional  stimuli  can  be  remembered  more;  simply  by  virtue  of  its  proximity  to  the  emotional  event.    According  to  Priority  Binding  Theory  (see  esp.  Mackay  et  al  2004  and  Mather  2007),  we  bind  congruent  information  together  using  semantic,  phonetic  and  visual  cues  to  create  a  cohesive  memory.    Memories  often  form  when  there  is  a  strong  semantic  or  contexual  association  (i.e.  table-­‐chair  or  bread-­‐eat)  between  the  emotional  target  stimuli  and  the  incidental  information  that  provides  a  possibility  for  binding  information  to  the  emotional  event  (Medford  et  al,  ).    Previous  studies  have  shown  that  during  negative  emotional  events,  there  can  be  a  memory  advantage  for  information  that  is  encoded  in  spatial  or  temporal  proximity  but  that  is  not  semantically  or  contexually  related  to  the  emotional  event    (Libkuman et al, 1999; Anderson et al 2006).  However,  to  our  knowledge,  this  has  never  been  demonstrated  under  experimental  conditions  for  information  encoded  in  positive  emotional  contexts.      Therefore  the  current  study  sought  to  investigate  whether  there  would  be  a  difference  in  memory  recognition  between  words  presented  with  positive  emotional  pictures  and  neutral  emotional  pictures.    The  current  study  paired  neutral  words  with  semantically  and  contextually  unrelated  pictures.    Therefore,  if  there  is  an  increase  in  recognition  memory  for  words  presented  with  positive  emotional  pictures  over  neutral  pictures,  this  may  suggest  that  there  is  a  memory  advantage  for  information  encoded  in  spatial  and  temporal  proximity  to  positive  emotional  events.    Alternatively,  if  there  is  a  decrease  in  recognition  memory  for  words  presented  with  positive  emotional  pictures,  this  would  suggest  that  the  positive  emotional  stimuli  is  interfering  with  the  processing  of  the  target  information.      Previous  studies  have  shown  that  reward  incentives  increase  performance  in  memory  tasks  for  pictures  (Adcock  et  al,  2006,  Whittman  et  al,  2005),  location  (Kawagoe  et  al,  1998),  words  (Madan  and  Spetch,  2012)  and  complex  semantic  information  (Marayame  &  Kuhbandner,  2011).    Neuroimaging  studies  have  demonstrated  that  reward  incentives  increase  memory  through  activation  of  both  lateral  prefrontal  areas  that  are  implicated  in  cognitive  control  (Watanabe,  1996,  Leon  &  Shadlen,  1999)  and  dopamine  pathways  (Berridge,  2007).    Reward  and  positive  affective  states  are  so  closely  related  that  a  number  of  theorists  suggest  that  they  are  entirely  indistinguishable.    For  instance,  Rolls  (2000)  argues  that  positive  emotion  is  a  bodily  experience  of  the  anticipation  or  attainment  of  goals  and  negative  emotion  is  the  bodily  experience  of  the  avoidance  of  threat.    It  is  surprising  then,  that  few  experiments  have  sought  to  investigate  the  interaction  between  reward  and  positive  effect  on  cognitive  functions  such  as  memory.    Two  recent,  previous  studies  have  explored  the  interaction  of  emotion  and  reward  on  memory  encoding.    Whitmann  et  al  (2008)  compared  memory  for  positive  emotional  stimuli  with  neutral  emotional  stimuli  under  reward  incentive.    They  

found  an  interaction  between  the  two  variables;  reward  enhanced  memory  significantly  more  in  the  positive  emotion  condition  than  in  the  neutral  emotion  condition.    However,  in  this  experiment,  participants  were  not  instructed  to  remember  the  stimuli,  so  encoding  was  incidental  and  secondly,  the  reward  incentive  was  not  related  to  memory  encoding  but  to  performance  in  an  immediately  ensuing  number  comparison  task.    A  second  study  by  Shigemune  et  al  (2010)  compared  memory  for  negative  emotional  stimuli  with  neutral  emotional  stimuli  under  reward  incentive.    As  expected,  they  found  that  negative  affective  stimuli  were  remembered  better  than  neutral  stimuli  and  that  reward  increased  memory  for  both  negative  and  neutral  stimuli  in  equal  measures,  i.e.  there  was  no  interaction  between  reward  and  negative  stimuli.      Therefore  the  present  study  sought  to  explore  whether  participants  incentivized  with  reward  would  perform  better  in  a  recognition  task.    Secondly,  if  there  is  a  memory  advantage  for  incidental  information  that  is  encoded  in  a  positive  emotional  or  neutral  context,  would  reward  have  an  additive  effect  on  memory  improvement  in  either  condition?    Unlike  Whitmann  et  al  (2008),  in  this  study  reward  was  related  to  success  in  the  memory  task  and  participants  were  instructed  to  remember  target  items,  so  encoding  was  intentional.    Thirdly,  if  there  is  a  memory  disadvantage  for  incidental  information  encoded  in  a  positive  emotional  context  (i.e.  if  emotional  stimuli  interfered  with  memory  encoding  for  proximate  information),  would  reward  reduce  the  interference  effect  through  an  increase  in  goal  motivation  and,  therefore,  an  increase  in  task  focus?        Method      Design    A  mixed  factorial  design  was  used  to  investigate  the  effect  of  emotional  context  and  reward  on  memory  for  target  words.    Specifically,  the  experiment  explored  whether  pairing  neutral  words  with  emotional  images  would  have  an  effect  on  performance  in  a  recognition  task  for  the  words  and  whether  promise  of  reward  would  increase  performance  in  the  task.  The  dependent  variable  was  recognition  memory,  measured  by  number  of  target  words  correctly  identified  in  a  recognition  task.    The  independent  variables  were  emotional  context  and  reward.    The  within-­‐subjects  factor  was  emotional  context,  with  two  levels;  words  presented  with  neutral  images  and  words  presented  with  positive  images.    The  between-­‐subjects  factor  was  reward,  with  two  levels;  reward  and  non-­‐reward.    This  was  manipulated  by  incentivising  the  reward  group  during  the  experiment  instructions.    The  experiment  was  conducted  using  a  web-­‐based  application  that  presented  50  randomly  ordered  and  randomly  paired  compound  stimuli:  neutral  target  words  presented  centrally  in  front  of  either  positive  emotional  images  or  neutral  images.  Following  a  maths  distractor  task,  participants  completed  a  yes/no/uncertain  recognition  task  to  test  memory  for  target  words.    

 Participants    107  participants  (57  females  and  50  males)  took  part  in  the  experiment.    This  was  a  convenience  sample  taken  from  colleagues,  extended  friends  and  family  of  the  researchers,  recruited  through  email  and  personal  invitation  to  take  part  in  a  ‘memory  experiment’.        Previous  studies  have  shown  a  progressive  decline  in  memory  function  with  ageing  (Castel  et  al  2002  &  Hedden  &  Gabrielii,  2004).    A  study  by  Warring  &  Kesinger  (2009)  also  revealed  that  adults  over  the  age  of  50  show  decreased  interference  effect  of  arousing  stimuli  in  memory  tasks.    Therefore,  recruitment  selection  ensured  that  all  participants  were  between  the  age  of  20  and  50.    Before  the  task,  participants  had  to  complete  an  online  form  that  required  an  entry  for  age.    Participants  could  not  progress  to  the  experiment  unless  they  were  between  20  and  50.    Further  control  for  age  was  achieved  through  the  application  counterbalancing  the  groups  for  age  across  a  median  split.  Mean  age  was  33.76  with  a  range  of  21-­‐50  and  a  standard  deviation  of  6.81.      Recruited  participants  all  spoke  English  to  a  high  level  of  fluency.    Following  the  completion  of  the  experiment,  box  plots  were  used  to  analyse  the  data-­‐sets  to  look  for  outliers  in  false  positive  scores.    Unusually  high  false  positive  scores  would  suggest  the  participant  had  either  guessed  responses  or  had  a  very  poor  memory.        Materials  /  Stimuli    25  neutral  and  25  positive  pictures  were  selected  from  the  International  Affective  Picture  System,  a  selection  of  a  standardised  images  developed  by  Lang  et  al  (2005).    The  pictures  have  been  previously  rated  on  a  10-­‐point  scale  for  both  arousal  and  valence.    Neutral  condition  pictures  had  a  mean  valence  of  5.07,  SD  =  0.34  and  mean  arousal  of  2.97,  SD  =  1.06.    Positive  condition  pictures  had  a  mean  valence  was  8.3,  SD  =  0.76  and  arousal  rating  of  4.52,  SD  =  1.34.    T-­‐tests  revealed  significant  differences  between  valence  t(48)  =  19.83,  p<0.001  and  arousal  t(48)  =  3.02,  p=0.007.        100  neutral  words  were  selected  from  Affective  Norms  for  English  Words  -­‐  ANEW  (Bradley  and  Lang,  1999).    The  mean  valence  for  the  words  was  4.98,  SD  =  0.17.    In  each  trial,  25  words  were  randomly  assigned  by  the  application  to  be  presented  with  positive  pictures  and  25  words  with  the  neutral  pictures.      A  further  50  words  were  presented  in  random  order  with  these  50  target  words  in  a  subsequent  recognition  task.      Reward  was  manipulated  by  embedding  within  the  task  instructions  a  promise  of  reward  (entry  into  a  £50  prize  draw)  for  high  performance  (a  score  >50%).    Care  was  taken  to  control  for  semantic  context,  which  has  been  shown  to  increase  levels  of  

processing  and  improve  performance  in  a  recall  task  (i.e.  Craik  &  Lockhart,  1972).      Associations  between  words  and  pictures  were  checked  separately  by  all  researchers  so  that  no  word  represented  any  object  presented  in  any  of  the  pictures.    Furthermore,  the  application  randomly  associated  words  and  pictures  across  participants.            Procedure      The  experiment  was  presented  as  a  web-­‐based  application  accessed  at  http://picturewords.herokuapp.com.  An  initial  instruction  page  informed  participants  that  the  experiment  investigates  visual  memory,  that  data  would  be  collected  anonymously  and  used  confidentially  and  that  they  could  retract  their  data  by  leaving  the  webpage  before  completion  of  the  experiment.      Participants  were  asked  to  ensure  they  were  in  a  quiet  environment  without  distraction  and  had  a  good  connection  to  the  web.    Participants  then  had  to  complete  a  web  form  that  recorded  age  (20  -­‐  50)  and  gender  (M  or  F).    Consent  was  gained  using  a  click  box,  which  bought  participants  onto  the  instruction  page.        Participants  were  instructed  that  they  would  be  shown  a  series  of  images  and  words  and  their  task  was  to  remember  the  words,  as  they  would  later  be  asked  to  identify  them  from  a  list.    The  reward  group  were  here  informed  of  a  reward  and  told  that  they  could  submit  their  email  address  at  the  end  of  the  experiment.    Immediately  after,  a  screen  was  presented  for  5  seconds  with  the  words  ‘please  focus  on  the  fixation  cross’.    A  fixation  cross  was  presented  concurrently,  50  pixels  below  the  centre  of  the  screen.        Following  this,  50  neutral  words  were  consecutively  presented  with  25  neutral  pictures  and  25  positive  pictures.    Each  randomly-­‐paired  stimuli  were  presented  for  3  seconds.    A  crosshair  was  presented  for  1  second  between  stimuli,  50  pixels  below  the  centre  of  the  screen.    Words  were  presented  50  pixels  below  the  centre  of  the  screen,  in  white,  against  a  black  background,  in  front  of  the  picture  (see  fig  1).    This  was  to  ensure  that  the  words  were  centrally  presented  but  did  not  obscure  key  information  in  the  pictures.          Following  presentation  of  stimuli,  participants  completed  eight  simple  arithmetic  problems.    This  was  a  distractor  task  to  control  for  recency  effect  and  to  ensure  that  words  were  not  retained  in  working  memory.      Participants  were  then  presented  with  a  second  instruction  page,  which  informed  participants  that  they  would  be  shown  a  series  of  100  words.    They  were  informed  that  they  should  press  "Y"  if  the  word  appeared  in  the  previous  task,  ‘N’  if  the  word  did  not  appear  and  “U”  if  they  were  uncertain.    At  this  stage,  those  assigned  to  the  reward  group  were  once  again  reminded  about  the  reward.    Words  were  presented  in  the  middle  of  the  screen,  with  the  above  instructions  directly  underneath  (see  fig  2).    Pressing  either  the  ‘Y’,  ‘N’  or  ‘U’  key  initiated  a  

progression  to  the  next  word.    A  progress  bar  ran  across  the  bottom  of  the  screen,  so  that  participants  could  chart  their  progress.    Following  the  randomly  ordered  presentation  of  all  50  target  words  and  50  novel  neutral  words,  participants  were  thanked  for  taking  part  and  reminded  of  the  anonymity  of  the  data  collection.      A  participant  number  was  presented  and  participants  were  informed  that  they  could  retract  their  data  by  sending  an  email  with  their  participant  number  to  a  given  email  address.    Finally,  those  in  the  reward  group  were  asked  to  enter  their  email  address  in  a  web  form,  if  they  wished  to  enter  the  £50  prize  draw.      Results    Following  analysis  of  error-­‐rates  using  boxplots,  participants  who  scored  outside  the  95%  confidence  interval  were  removed  from  the  final  analysis.    This  resulted  in  the  removal  of  the  data  of  5  participants,  leaving  102  participants  in  the  dataset.        A  repeated  measures  ANOVA  was  conducted  to  compare  the  effect  of  two  different  interventions,  (emotional  context  and  reward)  on  performance  in  a  word  recognition  task.    Results  are  presented  in  plot  1.    There  was  no  significant  main  effect  of  emotional  context  /  picture  category  on  the  number  of  words  recognised,  F(1,102)=0.773,  p=0.381,  with  no  difference  between  the  mean  number  of  words  recalled  in  the  positive  images  condition  (M=14.87,  SD=0.3)  and  the  neutral  images  condition  (M=14.32,  SD=0.3.    This  suggests  that  pairing  neutral  words  with  emotional  images  has  no  effect  on  memory  for  the  neutral  words.        There  was  no  significant  effect  of  reward  on  the  number  of  words  correctly  recognised,  F(1,102)=1.603,  p=0.208.    Although  results  fall  outside  of  significance,  they  fall  in  the  expected  direction.      The  mean  number  of  words  recalled  in  the  reward  condition  was  13.9,  SD=0.40.    The  mean  number  of  words  recalled  in  the  reward  condition  was  15.3.  SD=0.39.      However,  the  lack  of  significance  suggests  that  promise  of  reward  had  no  impact  on  performance  in  the  memory  task.              There  was  also  no  significant  interaction  between  the  picture  conditions  and  the  reward  condition.      F(1,102)  =  0.437,  p=0.510.      This  indicates  that  emotional  valence  does  not  mediate  the  effect  of  reward.        Further  analysis  of  the  data  was  conducted  in  order  to  control  for  participant’s  response  bias  and  task  accuracy.      Scores  were  standardised  as  z-­‐scores,  then  accuracy  was  indexed  by  calculating  hit  rates  minus  false  alarms  which  converted  scores  to  d’.    However,  results  still  remained  insignificant  for  each  variable  and  their  interaction.                Discussion    

No  significant  differences  were  found  at  test  between  recognition  scores  for  words  presented  with  positive  or  neutral  stimuli.    Memory  for  words  presented  with  emotional  stimuli  was  not  impaired,  in  comparison  with  words  presented  with  neutral  stimuli.    Therefore,  the  emotional  stimuli  presented  with  the  target  word  did  not  interfere  with  the  processing  of  their  compound  words.    The  lack  of  significance  in  the  experiment  results  may  be  due  to  an  artifact  of  the  experiment  design  or  it  may  point  to  some  more  interesting  suggestions  for  further  research.        Firstly,  it  is  possible  that  the  pictures  themselves  were  not  sufficient  to  induce  an  affective  response.    Although  these  photographs  have  been  previously  standardized,  many  were  taken  over  20  years  ago,  so  they  may  have  lost  cultural  relevance  and  therefore  their  ability  to  induce  affective  changes.    There  is  also  evidence  that  effects  of  emotion  on  memory  increase  with  time  (LaBar  &  Phelps,  1998).    It  is  possible  that  our  manipulation  would  have  become  more  evident  if  the  task-­‐test  interval  was  increased.    It  is  also  possible  that  our  simple  yes/no/uncertain  recognition  test  design  was  not  sensitive  enough  to  detect  the  differences  between  memory  for  neutral  information  presented  with  emotional  and  neutral  images.            Previous  experiments  have  found  that  emotional  stimuli  interfere  with  cognitive  processing  (Whalen  et  al.,  1998,    Kensinger  et  al  2007,).    There  may  be  a  number  of  reasons  why  the  present  study  did  not  find  this  effect.    Results  may  indicate  that  participants  were  able  to  focus  on  the  task  and  avoid  distraction  by  the  emotional  stimulus.    This  outcome  may  be  seen  as  a  confirmation  of  goal-­‐relevant  theories  of  emotional  memory  processing  (Levine  &  Edelstein,  2010)  over  against  Easterbrook’s  (1958)  Cue  Utilisation  hypothesis:  Participants  did  not  automatically  respond  to  the  presence  of  the  emotional  stimuli  but  were  able  eliminate  the  bottom-­‐up  stimulus-­‐response  effect  of  the  pictures  and  continue  with  top-­‐down  task-­‐  oriented  memory  encoding.    Experiments  using  the  emotional  stroop  task  (Whalen  et  al  1998)  have  found  that  clinical  populations  such  as  frontal  lobe  patients  exhibit  much  lower  thresholds  for  emotional  interference  (Bar-­‐Haim  et  al  2007).    It  may  be  possible  that  a  comparison  with  clinical  populations  against  the  general  population  sample  used  in  this  study  may  reveal  interference  effects  of  emotional  stimuli  on  memory  encoding.          Previous  experiments  that  found  an  increase  in  memory  encoding  for  neutral  information  in  proximity  to  emotion  stimuli.    However,  these  experiments  have  either  maintained  a  semantic  /  contextual  link  between  the  neutral  information  and  the  emotional  event    (for  instance  words  presented  in  emotional  sentences  -­‐  Brierley  et  al,  2007  and  Guillet  &  Arndt,  2010)  or  have  presented  the  information  alongside  arousing,  negative  affective  stimuli    (Libkuman  et  al  1990).          In  the  present  study  design,  care  was  taken  to  ensure  that  words  were  not  semantically  related  to  the  pictures  and  the  pictures  were  positive,  rather  than  negative  and  arousing.    Results  may  suggest  that  for  emotional  events  to  increase  memory  for  proximate  information,  the  information  needs  to  be  contextually  or  

semantically  related  to  the  event  and  /  or  needs  to  be  placed  proximately  to  highly  arousing,  negative  stimuli.        In  order  to  explore  this  further,  the  present  study  could  be  repeated  with  two  further  conditions:  neutral  words  presented  with  semantically  related  emotional  pictures  (i.e.  the  word  ‘bottle’  with  a  picture  of  a  baby)  and  neutral  words  presented  with  arousing,  negative  images.          We  may  hypothesize  that  neutral  words  presented  in  the  ‘semantically-­‐related’  condition  will  be  better  remembered  than  words  presented  with  semantically  unrelated  pictures.    This  is  likely  to  occur  because  the  contextual  /  semantic  relationship  between  the  stimuli  allows  the  word  to  be  bound  with  the  emotional  event,  enabling  the  memory  advantage  for  the  words,  seen  in  other  experiments.      Secondly,  we  may  hypothesize  that  words  presented  in  the  ‘negative  arousing’  condition  will  be  remembered  better  than  words  presented  with  semantically  unrelated  positive  images.      Maratos  et  al  (2001)  used  fMRI  to  investigate  the  different  areas  of  the  brain  involved  in  processing  information  encoded  in  positive  affective  and  neutral  affective  states.    They  found  that  information  encoded  in  negative  emotional  states  was  associated  with  activity  in  the  right  dorsolateral  prefrontal  cortex  as  well  as  activity  in  limbic  areas  such  as  the  amygdala  and  hippocampus.    By  contrast,  information  encoded  during  positive  emotional  states  was  associated  with  activity  in  bilateral  frontal  and  orbitofrontal  areas,  with  less  activity  in  limbic  areas.        This  may  provide  evidence  for  the  idea,  suggested  by  Levine  and  Burgess,  (1997)  and  Levine  and  Pizarro  (2004)  that  different  information  processing  strategies  are  associated  with  different  affective  events.    Learning  through  positive  emotion  requires  semantic  associations  to  be  established,  or  at  least  that  positive  emotions  facilitate  connective  thinking  and  assimilation  of  information  to  current  schema  information  networks.    In  contrast,  learning  through  negative  emotions  does  not  necessarily  need  semantic  associations;  temporal  and  spatial  proximity  is  sufficient  to  improve  memory  encoding  (see  Doerksen  and  Shimamura  (2001)  &  Anderson  et  al  (2006).    These  two  different  information  and  memory  encoding  strategies  have  obvious  evolutionary  benefits.      During  positive  emotional  experiences  we  broaden  our  thinking  to  seek  out  associations  between  the  positive  affect  and  environment.    Once  these  semantic  associations  are  established,  they  can  be  encoded  into  memory.    If  they  are  not  found,  peripheral  information  is  discarded.  However,  in  order  to  survive  threats,  early  primates  needed  to  respond  to  the  widest  possible  source  of  danger.    Therefore  anything  associated  with  the  source  of  threat  may  be  encoded  alongside  it  in  memory.    Further  studies  are  needed  to  explore  this  idea;  particularly  those  that  investigate  what  type  of  information  is  encoded  in  memory  during  different  affective  states.        No  significant  differences  were  found  between  the  reward  and  the  non-­‐reward  group.      In  light  of  overwhelming  previous  evidence  for  the  effect  of  reward  on  memory  tasks  (Adcock  et  al,  2006,  Whittman  et  al,  2005),  it  is  probable  that  our  

intervention  failed  to  incentivise  those  in  the  reward  group.    There  are  a  number  of  possible  reasons  for  this,  including  participants  calculating  that  it  is  unlikely  that  they  would  win  the  reward.      Alternatively,  it  may  be  that  because  all  participants  volunteered  to  take  part,  this  altruistic  act  in  itself  was  enough  of  a  reward  for  both  conditions,  that  the  intervention  itself  failed  to  have  any  additive  effect.    Finally,  most  previous  experiments  that  have  found  significant  effects  for  reward  have  used  a  within  groups  design  for  reward.    For  instance  (Shigemune  et  al  2010)  presented  a  ‘reward  indicator’  before  a  reward  stimuli  is  presented,  so  that  the  difference  between  a  reward  stimulus  and  a  non-­‐reward  stimulus  was  constantly  reinforced.    Finally  there  was  no  observed  interaction  between  reward  and  emotional  context.      Reward  was  not  observed  to  have  an  additive  effect  on  memory  ability  for  information  encoded  in  either  positive  or  neutral  conditions,  despite  the  intentionality  of  the  encoding  and  the  association  between  the  task  and  the  reward  incentive.    As  there  was  no  memory  disadvantage  for  incidental  information  encoded  in  a  positive  emotional  context  results  did  not  confirm  whether  reward  incentives  would  reduce  the  interference  effect  of  emotional  stimuli.    In  order  to  explore  this  further,  we  would  need  to  ensure  that  our  reward  intervention  was  effective.    Secondly,  it  would  probably  be  more  effective  to  explore  the  reward  /  affect  interaction  by  comparing  memory  for  emotional  pictures  versus  neutral  pictures,  rather  than  testing  information  encoded  in  proximity  to  emotional  pictures,  as  occurred  during  the  present  experiment.        To  conclude,  the  current  study  found  that  by  placing  neutral  affective  words  with  positive  images  neither  increased  nor  decreased  performance  in  a  subsequent  word  memory  task.      The  study  also  found  no  effect  of  a  reward  intervention  on  performance  in  the  recognition  task  or  an  interaction  between  the  variables.    Results  may  be  due  to  experimental  artifacts,  may  indicate  that  task  orientation  mediates  the  effect  of  emotional  interference  or  may  indicate  that  positive  emotional  contexts  only  have  an  additive  effect  on  memory  when  the  information  can  be  bound  with  the  emotional  event  through  contextual,  semantic  or  pragmatic  associations.    However,  this  final  conclusion  is  tentative  and  requires  further  investigation.                                

         References        Adcock,  R.  A.,  Thangavel,  A.,  Whitfield-­‐Gabrieli,  S.,  Knutson,  B.,  &  Gabrieli,  J.  D.  E.  (2006).  Reward-­‐motivated  learning:  Mesolimbic  activation  precedes  memory  formation.  Neuron,  50,  507–517.    Anderson,  A.K.,  Wais,  P.E.,  &  Gabrieli,  J.D.E.  (2006).  Emotion  enhances  rememberance  of  neutral  events  past.  Proceedings  of  the  National  Academy  of  Sciences,  103,  1599-­‐1604.    Bar-­‐Haim,  Y.,  Lamy,  D.,  Pergamin,  L.Bakermans-­‐Kranenberg,  M.J.,  &  van  Ijzendoorn,  M.H.  (2007).  Threat  related  attentional  bias  in  anxious  and  nonanxious  individuals:  A  meta-­‐analystic  study.  Psychological  bulletin,  133,  1,  1-­‐24.    Berridge,  K.C.  (2007).  The  debate  over  dopamine's  role  in  reward:  the  case  for  incentive  salience.  Psychopharmacology,  191,  391-­‐431.    Bradley,  M.M.,  &  Lang,  P.  J.  (1999).  Affective  norms  for  English  words  (ANEW):  Stimuli,  instruction  manual  and  affective  ratings  (Technical  report  No.  C-­‐1).  In  F.  L.  Gainesville  (Ed.),  The  Center  for  Research  in  Psychophysiology:  University  of  Florida.    Bridge,  D.  J,  Chiao,  J.  Y.,  &  Paller,  K.  A.  (2010).  Emotional  context  at  learning  systematically  biases  memory  for  facial  information.  Memory  &  Cognition,  38,  125–133.    Brierley,  B.,  Medford,  N.,  Shaw,  P.,  &  David,  A.  (2007).  Emotional  memory  for  words:  Separating  content  and  context.    Cognition  and  Emotion,  21,  495-­‐521.    Castel,  A.  D.,  Benjamin,  A.  S.,  Craik,  F.  I.  M.,  &  Watkins,  M.  J.  (2002).  The  effects  of  aging  on  selectivity  and  control  in  short-­‐term  recall.  Memory  &  Cognition,  30,  1078–1085.    Christianson,  S.A.,  &  Lotus,  E.  F.  (1991).  Remembering  emotional  events:  The  fate  of  detailed  information.  Emotion  &  Cognition,  5,  81-­‐108.    Craik,  F.;  Lockhart,  R.  (1972).  Levels  of  processing:  A  framework  for  memory  research.  Journal  of  Verbal  Learning  and  Verbal  Behaviour,  11,  6,  671–684    Doerksen,  S.,  &  Shimamura,  A.P.  (2001)  Source  Memory  Enhancement  for  Emotional  Words.  Emotion,  1,1,  5-­‐11.      

Easterbrook,  J.A.  (1959).  The  effect  of  emotion  on  cue  utilization  and  the  organization  of  behavior.  Psychological  Review,  66,  183-­‐201.    Fredrickson,  B.  L.  (2001).  The  role  of  positive  emotions  in  positive  psychology:  The  broaden-­‐and-­‐build  theory  of  positive  emotions.  American  Psychologist,  56,  218-­‐226.    Gootjes,  L.,  Coppens,  L.C.,  Zwaan,  R.A.,  Franken,  I.H.,  Van  Strien,  J.,  &  Wint,  J.  (2011).    Effects  of  recent  word  exposure  on  emotion-­‐word  Stroop  interference:  an  ERP  study.  Psychophysiology,  79,  3,  356-­‐63.    Guillet,  R.  &  Arndt,  J.  (2009).  Taboo  words:  The  effect  of  emotion  on  memory  for  peripheral  information.  Memory  &  Cognition,  37,  866-­‐879        Hamann,  S.B.,  Ely,  T.D,  Grafton,  S.T.,  &  Kilts,  C.D.  (1999).  Amygdala  activity  related  to  enhanced  memory  for  pleasant  and  aversive  stimuli.  Nature  Neuroscience,  2,  289  –  293.    Hedden  T,  Gabrielii  J.D.  (2004)  Insights  into  the  ageing  mind:  a  view  from  cognitive  neuroscience.  Nature  Reviews  Neuroscience.  5,  2,  87-­‐96.    Kawagoe,  R.,  Takikawa,  Y.,  &  Hikosaka,  O.  (1998).  Expectation  of  reward  modulates  cognitive  signals  in  the  basal  ganglia.  Nature  Neuroscience.  1,  5,  411-­‐416    Kensinger,  E.A.,  &  Corkin,  S.  (2003).  Memory  enhancement  for  emotional  words:  Are  emotional  words  more  vividly  remembered  than  neutral  words?  Memory  &  Cognition,  31,  1169  –  1180.    Kensinger,  E.A.,  Garoff-­‐Eaton,  R.J.,  &  Schacter,  D.L.  (2007).  Effects  of  emotion  on  memory  specificity:  memory  trade-­‐offs  elicited  by  negative  visually  arousing  stimuli.  Journal  of  Memory  and  Language,  56,  575-­‐591.      LaBar,  K.S.,  &  Phelps,  E.A.  (1998).  Arousal-­‐mediated  memory  consolidation:  Role  of  the  medial  temporal  lobe  in  humans.  Psychological  Science,  9,  490  –  493.    Lang,  P.J.,  Bradley,  M.M.,  &  Cuthbert,  B.N.  (2005).  International  affective  picture  system  (IAPS):  Affective  ratings  of  pictures  and  instruction  manual.  Technical  Report  A-­‐6.  University  of  Florida,  Gainesville,  FL.    Leon,  M.I.,  &  Shadlen  M.N.  (1999).  Effect  of  expected  reward  magnitude  on  the  responseof  neurons  in  the  dorsolateral  prefrontal  cortex  of  the  macaque.  Neuron,  24,  415-­‐425.    Levine,  L.J.  &  Burgess,  S.L.  (1997).  Beyond  general  arousal:  Effects  of  specific  emotions  on  memory.  Social  Cognition,  15,  3  157-­‐181.    Levine  L.J.  and  Pizarro  D.A.  (2004)  Emotion  and  memory  research:  A  grumpy  

overview.  Social  Cognition,  22,  5,  530-­‐554.    Levine,  L.J.  &  Edelstein,  R.S.  (2010).  Emotion  and  memory  narrowing:  A  review  and  goal-­‐relevance  approach.  In  De  Houwer,  J.  &  Hermans,  D.  (eds).  Cognition  and  Emotion.  Psychology  Press,  168-­‐210.      Libkuman,  T.  M.,  Nichols-­‐Whitehead,  P.,  Griffith,  J.,  &  Thomas,  R.  (1999).  Source  of  arousal  and  memory  for  detail.  Memory  and  Cognition,  27,  1,  166-­‐190.        MacKay,  D.  G.,  Shafto,  M.,  Taylor,  J.  K.,  Marian,  D.  E.,  Abrams,  L.,  &  Dyer,  J.  R.  (2004).  Relations  between  emotion,  memory,  and  attention:  Evidence  from  taboo  Stroop,  lexical  decision,  and  immediate  memory  tasks.  Memory  &  Cognition,  32,  474–488.    Madan,  C.R.,  &  Spectch  M.L.  (2012).  Is  the  enhancement  of  memory  due  to  reward  driven  by  value  or  salience?  Acta  Psychologica,  139,  343–349.    Maratos,  E.J.,  Dolan,  R.J.,  Morris,  J.S.,  Henson,  R.N.,  Rugg,  M.D.  (2001)  Neural  activity  associated  with  episodic  memory  for  emotional  context.  Neuropsychologia,  39,  910-­‐920.    Mather  M.  (2007).  Emotional  arousal  and  memory  binding.  Perspectives  on  Psychological  Science,  2,  33–52.    Mather,  M.,  &  Nesmith,  K.  (2008).  Arousal-­‐enhanced  location  memory  for  pictures.  Journal  of  Memory  and  Language.  58,  449-­‐464.    Medford,  N.,  Phillips,  M.L.,  Brierley,  B.,  Brammer,  M.,  Bullmore.  E.T.,  &  David,  A.S.  (2005).  Emotional  memory:  Separating  content  and  context.  Psychiatry  Research:  Neuroimaging,  138,  247–258    Murayama  ,  K.,  &  Kuhbandner,  C.  (2011).  Money  enhances  memory  consolidation  –  But  only  for  boring  material.  Cognition,  119,  120-­‐124.      Ochsner,  K.N.  (2000).  Are  affective  events  richly  recollected  or  simply  familiar?  The  experience  and  process  of  recognizing  feelings  past.  Journal  of  Experimental  Psychology,  129,  2,  242  -­‐  61.    Rolls,  E.T.  (2000).  Précis  of  the  brain  and  emotion.  Behavioral  and  Brain  Sciences,  23,  2,  177-­‐191.  

Shigemune,  Y.,  Abe,  N.,  Suzuki,  M.,  Ueno,  A.,  Mori,  E.,  Tashiro,  M.,  et  al.  (2010).  Effects  of  emotion  and  reward  motivation  on  neural  correlates  of  episodic  memory  encoding:  A  PET  study.  Neuroscience  Research,  67,  72–79.    Siegrist,  M.  (1995).  Effects  of  taboo  words  on  color-­‐naming  performance  on  a  Stroop  test.  Perceptual  and  Motor  Skills,  81,  1119-­‐1122.      

   Talirico,  J.M.,  &  Rubin,  D.C.  (2003).  Confidence,  not  consistency,  characterizes  flashbulb  memories.  Psychological  Science,  14,  455-­‐461.    Talirico,  J.M.,  Berntsen,  D.,  &  Rubin,  D.C.  (2009).  Postiive  emotions  enhance  the  recall  of  peripheral  details.    Cognition  and  Emotion,  23,  380-­‐398.    Tsukiura,  T.  &  Cabeza,  R.  (2008).  Orbitofrontal  and  hippocampal  contributions  to  memory  for  face-­‐name  associations:  the  rewarding  power  of  a  smile.  Neuropsychologia,  46,  2310-­‐2319.  

Waring,  J.D.,  &  Kensinger,  E.A.  (2009).  Effects  of  Emotional  Valence  and  Arousal  Upon  Memory  Trade-­‐Offs  With  Aging.    Psychology  and  Aging,  24,  2,  412–422.    Watanabe,  M  (1996).  Reward  expectancy  in  primate  prefrontal  neurons.  Nature,  382,  629-­‐632.      Whalen,  P.  J.,  Bush,  G.,  McNally,  R.  J.,  Wilhelm,  S.,  McInernery,  S.  C.,  Jenike,  M.  A.,  and  Rauch,  S.  L.  (1998).  The  emotional  counting  stroop  paradigm:  A  functional  magnetic  resonance  imaging  probe  of  the  anterior  cingulate  affective  division.  Biological  Psychiatry,  44,  1219–228.    Wittmann,  B.  C.,  Schott,  B.  H.,  Guderian,  S.,  Frey,  J.  U.,Heinze,  H.  J.,  &  Düzel,  E.  (2005).  Reward-­‐related  fMRI  activation  of  dopaminergic  midbrain  is  associated  with  enhanced  hippocampus-­‐dependent  long-­‐term  memory  formation.  Neuron,  45,  459–467.    Whitmann,  B.C.,  SChiltz,  K.,  Boehler,  C.N.,  Duzel,  E.  (2008).  Mesolimbic  interaction  of  emotional  valence  and  reward  improves  memory  formation.    Neuropsychologia,  46,  1000-­‐1008                                  

         Appendix    Fig  1.  Compound  Stimuli  Example    

         Fig  2.  Recognition  Test  Example    

   

 Butter  

 Was  this  word  included  in  task  1?  

 Press  ‘Y’,  ‘N’  or  ‘U’.  

   

     Plot  1.