The defendant, David Walters, has been charged with the aggravated felonious sexual assault of his minor stepdaughter based on conduct that allegedly occurred between August 1984 and November 1985. See RSA 632-A:2, XI (Supp. 1983). This interlocutory appeal from ruling, see SUP. CT. R. 8, requires us to determine the admissibility of a complaining witness’s testimony in a sexual assault trial when the witness, for a period of time, did not remember the charged conduct. See State v. Hungerford, 142 N.H. 110, 697 A.2d 916 (1997). The Superior Court (Dalianis, J.) denied the defendant’s motion to suppress the evidence, ruling that the memory must be assumed to be no less reliable than ordinary human memory. We reverse and remand.
At an evidentiary hearing, the complainant and her therapist testified to the following facts. The complainant, then fifteen years old, became pregnant in August 1991. She informed her mother of her pregnancy in December of that year. At some point after informing her mother but before giving birth, the complainant had three “nightmares,” during which she had “flashbacks” which consisted of glimpses of the defendant abusing her when she was nine or ten years old. See Ernsdorff & Loftus, Let Sleeping Memories Lie? Words of Caution About Tolling the Statute of Limitations in Cases of Memory Repression, 84 J. CRIM. L. & CRIMINOLOGY 129, 138 (1993) (describing recovery of repressed memory via a flashback, or “a reliving of a traumatic experience as if it were currently happening”). Before having these nightmares, the complainant apparently had no memory of being assaulted by the defendant.
Although she could not recall at the hearing in what order she had the three nightmares, the complainant described their content as follows:
The first one that I can recall was when [the defendant] put me to bed and he pulled the sheet over, that was — the sheet off [of] me and that was the only thing that was in the first flashback. The second one was — it wasn’t clear to me at the time but it was when [the defendant] was down *241between my legs. The third one was when the next mornin’ it seemed everything was fine. . . . The next morning of the night that it happened. The following morning was the last flashback.
Upon questioning by defense counsel, the complainant testified that the dreams themselves did not “put everything together” about the defendant’s conduct:
Question: You said you put your memories together. You did that after you had woken up again?
Answer: No, I just put the flashbacks together, not when I woke up but after I had the third one I put ’em together and I realized that I was molested by [the defendant].
Question: I’m trying to get how you put them together. Was your third flashback a longer flashback than the others?
Answer: No.
Question: So it was another chunk?
Answer: Yes.
Question: And then I guess - did the realization hit you, after you had woken up?
Answer: Yes.
Question: So you woke up, remembered the nightmare and ended up thinking about it more?
Answer: Yes.
The complainant was engaged in psychological therapy before, during, and after the nightmares. She testified that she did not report the abuse to her therapist when she remembered it because she knew that her therapist would report any abuse to authorities and because she did not want to cause the breakup of her mother’s marriage to the defendant.
The complainant informed her boyfriend of her memory sometime before the birth of her child in May 1992. She declined to tell her mother about the memory until November 1992, again stating that she “didn’t want to break up [her] marriage” to the defendant. She revealed the memory to authorities at her school and to her therapist in April 1993. The instant charges followed.
The defendant moved to exclude the complainant’s testimony from being admitted at his trial, arguing that she should not be allowed *242to testify regarding her alleged memory of events that she previously had not remembered. The parties stipulated that the court could consider the record of the pretrial admissibility heai’ing on recovered memories in the two cases reported in Hungerford, 142 N.H. 110, 697 A.2d 916, which was pending before a different judge in the superior court. After considering the expert testimony from the Hungerford hearing and conducting a brief hearing on the facts of the instant case, the trial court ruled that the complainant’s testimony was admissible, and that the testimony was not subject to the threshold reliability requirements of expert testimony. The court found that the purported memories were not “the product” of the complainant’s therapy, based upon the court’s observation that the nightmares occurred during a period when her visits to therapy were infrequent and that “nothing in the therapy sessions concerned sexual abuse.” The court stated that “[bjecause the objective truth of memory cannot be scientifically determined, lay testimony of allegedly recovered memories cannot be barred on the basis that there has been no preliminary showing of its reliability.” The court ultimately ruled that expert testimony would be allowed to explain the phenomena of traumatic amnesia and memory repression, and to opine on the reliability of ordinary memory and recovered memory, but not to opine on the truth of a particular recovered memory. This interlocutory appeal followed.
The court is presented with nine questions in the interlocutory appeal statement, but, in light of our decision in Hungerford, 142 N.H. 110; 697 A.2d 916, we need only address two: whether the trial court erred in placing the burden of proof on the defendant to demonstrate the unreliability of recovered memories, and whether the court erred in ruling that the testimony was admissible. We defer to the trial court’s rulings on evidentiary matters and reverse only when the court has abused its discretion. State v. Briere, 138 N.H. 617, 620, 644 A.2d 551, 554 (1994). We also review the trial court’s decisions on the reliability of scientific evidence deferentially, “although we review some aspects of scientific evidence independently when their reliability or general acceptance is not likely to vary according to the circumstances of a particular case.” Hungerford, 142 N.H. at 117, 697 A.2d at 920; see State v. Vandebogart (DNA), 136 N.H. 365, 376, 616 A.2d 483, 491 (1992).
We agree with the defendant that the trial court improperly shifted the burden of proof in ruling on the defendant’s motion to suppress. The party offering evidence generally bears the burden of demonstrating its admissibility. See, e.g., Opinion of the Justices *243(Prior Sexual Assault Evidence), 141 N.H. 562, 577, 688 A.2d 1006, 1015 (1997); State v. Coleman, 133 N.H. 713, 715, 584 A.2d 755, 757 (1990). In Hungerford, we confirmed that this general rule applies in the context of recovered memories; specifically, upon the objection of the opposing party, the proponent of testimony comprised of recovered memories must demonstrate that the testimony is reliable. Hungerford, 142 N.H. at 119, 697 A.2d at 921-22. The trial court erred in placing the burden on the defendant to demonstrate that recovered memories are not reliable. See id.
We turn to the admissibility of the complainant’s memories in the instant case. In Hungerford, we concluded that recovered memories must satisfy a threshold reliability inquiry to be admitted. Id. at 118, 697 A.2d at 921. We ruled that “[t]o establish that a recovered memory is reliable, the proponent of its admission must demonstrate a reasonable likelihood that the recovered memory is as accurate as ordinary human memory.” Id. at 125, 697 A.2d at 924. We listed eight factors that the court should consider in ruling on the admissibility of this testimony:
(1) the level of peer review and publication on the phenomenon of repression and recovery of memories; (2) whether the phenomenon has been generally accepted in the psychological community; (3) whether the phenomenon may be empirically tested; (4) the potential or known rate of recovered memories that are false; (5) the age of the witness at the time the event or events occurred; (6) the length of time between the event and the recovery of the memory; (7) the presence or absence of objective, verifiable corroborative evidence of the event; and (8) the circumstances attendant to the witness’s recovery of the memory, i.e., whether the witness was engaged in therapy or some other process seeking to recover memories or likely to result in recovered memories.
Id. at 125-26, 697 A.2d at 925 (citations omitted).
The trial court considered the same expert testimony in this case that we reviewed in Hungerford. The first four factors of our reliability inquiry in Hungerford, accordingly, apply equally in the instant case, insofar as we inquire into the phenomenon of recovered memories itself. See id. at 117, 697 A.2d at 920 (supreme court reviews independently elements of reliability inquiry unlikely to change from case to case).
In Hungerford, we first considered the level of peer review and publication on the phenomenon of recovered memories and inquired *244into the question whether the phenomenon has been generally-accepted in the psychological community. Id. at 125, 697 A.2d at 925. We concluded that, although the level of peer review and publication on the phenomenon was quite high, id. at 126, 697 A.2d at 925, we could not conclude that there was general acceptance of the phenomenon in the psychological community, id. at 130, 697 A.2d at 928.
On the third consideration, whether it is possible to test the phenomenon empirically, we observed that “it would be impossible, ethically, to test repression and recovery of memory of severely traumatic events in a laboratory setting.” Id. at 130, 697 A.2d at 928. We also concluded that there was no real way to track the percentage or number of recovered memories that are “false,” especially when the phenomenon remains the subject of such debate. Id. at 131, 697 A.2d at 928. Our evaluation of these four factors of the inquiry weighed heavily against admitting the evidence in that case. See id. at 132, 697 A.2d at 930.
We turn to the factors of the reliability inquiry that relate to the particular recovered memory in the instant case. According to the indictments, the complainant was either eight or nine years old when the charged acts occurred in 1984 or 1985. She had the nightmares that led to the recovery of her memories at some point during her pregnancy, in 1992; accordingly, the complainant did not remember the conduct for seven or eight years after it occurred. As we observed in Hungerford, “[c]hildren who are very young are perceived to have incomplete narrative memories even of traumatic events.” Id. at 132, 697 A.2d at 928; see Pezdek & Roe, Memory for Childhood Events: Hoiv Suggestible is It?, 3 CONSCIOUSNESS & COGNITION 374, 375-76 (1994). The complainant’s youth at the time of the events weighs somewhat against the reliability of the memory. See Hungerford, 142 N.H. at 132, 697 A.2d at 929. Further, because of the innumerable external influences that may affect memory while it is in the retention stage, see Ernsdorff & Loftus, supra at 156-58, “a shorter period of time between the event and recall offers less opportunity for suggestion,” Hungerford, 142 N.H. at 132, 697 A.2d at 929. See, e.g., Hall et al., Postevent Information and Changes in Recollection for a Natural Event, in EYEWITNESS Testimony: Psychological perspectives 124, 132-38 (G. Wells & E. Loftus eds., 1984). Seven or eight years is a longer period of time during which the complainant had no memory of the event than in the two cases we considered in Hungerford. Hungerford, 142 N.H. at 132, 697 A.2d at 928-29.
*245On the record before us, we discern no objective, verifiable corroborative evidence that the charged acts occurred. See id. at 125-26, 697 A.2d at 929. We accordingly do not consider this prong in deciding the transferred questions in this appeal.
Finally, we consider the circumstances attendant to the recovery of the memory. See id. at 126, 697 A.2d at 929. Although the complainant was participating in psychological therapy at the time she recovered her memory, we accept the trial court’s finding that “the complainant’s memory of sexual abuse was not a product of therapy.” But see Taub, The Legal Treatment of Recovered Memories of Child Sexual Abuse, 17 J. LEGAL MED. 183, 191 (1996) (patient may recover memory in response to perceived expectation of therapist). We also consider the infrequency of the complainant’s visits during this period as diminishing the potential for suggestiveness inherent in the relationship. See, e.g., Loftus, The Reality of Repressed Memories, 48 Am. PSYCHOLOGIST 518, 526-33 (1993).
We briefly address the fact that the complainant recovered her memory by piecing together fragments that came to her as “flashbacks” during several dreams, rather than spontaneously or during a therapy session. See Ernsdorff & Loftus, supra at 138. The State does not argue that the content of the complainant’s flashbacks constitute the actual memories; rather, the State argues that she now possesses an independent recollection of the events that was spurred by the dreams. Our concerns with the suggestibility of memory, expressed in Hungerford, are equally applicable in the context of dreams. We observed:
Studies indicate that memory is not a mechanism that merely reproduces one’s perceptions of events; rather, memory, like perception, is an active, constructive process that often introduces inaccuracies by adding details not present in the initial representation or in the event itself. The mind combines all the information acquired about a particular event into a single storage “bin,” making it difficult to distinguish what the witness saw originally from what she learned later.
Hungerford, 142 N.H. at 124, 697 A.2d at 924 (quotation omitted). Further, we are concerned with the difficulty that the opponent of evidence from a dream-recovered memory would face in the event this sort of evidence is admitted; we can imagine the difficulty of cross-examining a dream to expose its weaknesses or flaws. Cf. State v. Cressey, 137 N.H. 402, 410, 628 A.2d 696, 701 (1993) (considering difficulty of effective cross-examination of expert witness where *246several interpretive steps made inside head of expert); Comment, Repression, Memory, and Suggestibility: A Call for Limitations on the Admissibility of Repressed Memory Testimony in Sexual Abuse Trials, 66 U. COLO. L. REV. 477, 505-06 (1995) (observing that some flashbacks may be “obsessional images” rather than memories).
On the basis of the record before us, we conclude, as we did in Hungerford, that “[t]he indicia of reliability present in the particular memories in [this] case[] do not rise to such a level that they overcome the divisive state of the scientific debate on the issue.” Id. at 134, 697 A.2d at 930; see also Note, Recovered Memories of Childhood Sexual Abuse: Applying the Daubert Standard in State Courts, 69 S. CAL. L. REV. 855, 859-67 (1996) (describing spectrum of scientific positions on question of reliability of recovered memories). We remand the case for futher proceedings consistent with this opinion, which may include reapplication of the Hungerford test if a significant change in available facts makes this appropriate.
Reversed and remanded.
HORTON, J., did not sit; the others concurred.